The defendants were convicted of involuntary manslaughter of their daughter, Rita Michaud. G. L. c. 265, § 13. On appeal, they claim error in the allowance of the Commonwealth’s motion to amend the indictments, and in the denial of their motions to dismiss the indictments and for required findings of not guilty. The Appeals Court affirmed the convictions. Commonwealth *492v. Michaud, 14 Mass. App. Ct. 471 (1982). We allowed the defendants’ applications for further appellate review. We conclude that the defendants’ motions for required findings of not guilty should have been allowed. Accordingly, the convictions are reversed.2
In reviewing the denial of the defendants’ motions for required findings of not guilty, we consider “‘whether the evidence, in its light most favorable to the Commonwealth . . . [was] sufficient ... to permit the jury to infer the existence of the essential elements of [involuntary manslaughter].’ Commonwealth v. Sandler, 368 Mass. 729, 740 (1975). Accord, e.g., Commonwealth v. Dunphy, 377 Mass. 453, 455-456 (1979); Commonwealth v. Seay, 376 Mass. 735, 737 (1978); Commonwealth v. Campbell, 375 Mass. 308, 311-312 (1978) . . . . [T]he evidence and the inferences permitted to be drawn therefrom must be ‘of sufficient force to bring minds of ordinary intelligence and sagacity to the persuasion of [guilt] beyond a reasonable doubt,’ as required by Commonwealth v. Cooper, [264 Mass. 368, 373 (1928)].” Commonwealth v. Latimore, 378 Mass. 671, 676-677 (1979). Commonwealth v. Borans, 379 *493Mass. 117, 134-135 (1979). Accord, Commonwealth v. Barrett, 386 Mass. 649, 655 (1982), and cases cited.
The evidence, in the light most favorable to the Commonwealth, is briefly summarized. Carol gave birth to Rita on June 29, 1980. Carol’s obstetrician testified that the pregnancy was without complications. At birth, the baby weighed seven pounds, six ounces, and appeared to be doing very well. Carol breast fed the baby in the hospital. Carol and the baby were discharged from the hospital on July 1, 1980, Rita then weighing six pounds, thirteen ounces. Carol’s obstetrician testified that it is normal for a baby to lose ten per cent of its weight in the first few days of life.
On July 24, 1980, at 9:43 a.m., Carol called the Blackstone police and requested emergency aid because her baby was not breathing. The first officer to arrive found Normand giving artificial respiration to the baby, and observed a small exchange of air in the baby. Shortly after that, observing no exchange of air, he started mouth-to-mouth resuscitation. An emergency medical technician arrived, and continued the resuscitative efforts until the baby arrived at the Fogarty Memorial Hospital in North Smifhfield, Rhode Island, where the emergency room staff took over. All efforts were unavailing and the baby was pronounced dead at 10:12 a.m., by Dr. Charles Mead, the emergency room physician.
It was Dr. Mead’s opinion that the baby was dead when she arrived at the emergency room, because she was cold, stiff, had no pulse, and was discolored. He testified that her condition was consistent with her having been deprived of oxygen for some period greater than fifteen minutes. He felt that the child looked undernourished, based on her size and lack of subcutaneous tissue, or baby fat. He stated that the nurses commented that “this child certainly looks very thin.” The Commonwealth introduced four photographs which Dr. Mead testified were a fair and accurate representation of the baby’s appearance when he examined her on July 24.
Dr. William Sturner, the chief medical examiner for the State of Rhode Island, performed an autopsy on the baby’s *494body commencing at approximately 12:30 p.m., on July 24. The Commonwealth introduced five photographs which Dr. Sturner testified fairly and accurately depicted the baby’s appearance just prior to the autopsy. In these photographs, the baby’s ribs and backbone are visible. Dr. Sturner testified that he found the baby’s body to be thin and malnourished. He based his opinion on the baby’s weight, which was approximately six pounds; the differential between that weight and the birth weight, amounting to a loss of approximately one pound, six ounces; and the baby’s prominent rib markings, and sunken cheeks and eyes due to the lack of subcutaneous tissue. There were no injuries or bruises. The baby’s diaper was soiled with feces and a moisture consistent with urine. Dr. Sturner found no evidence of abnormalities of the face or mouth area, and the autopsy revealed no evidence of abnormalities of the gastrointestinal tract or other internal organs which would prevent the absorption of food. The stomach was empty, but there were yellow mucous and fecal material in the distal colon and intestinal tract, indicating some ingestion in the preceding twenty-four to forty-eight hours, but not in the last four to six hours. He found no evidence other than starvation to account for the baby’s death, and in particular made a specific determination that the death was not caused by the sudden infant death syndrome, or crib death. On the basis of the autopsy, as well as several tests, it was Dr. Sturner’s opinion that the baby died of electrolyte imbalance due to malnutrition and dehydration as a result of starvation. He defined starvation as the lack of adequate or appropriate or sufficient sustenance and fluid getting to the tissues of the baby to sustain normal growth and development.
The Commonwealth introduced Carol’s and Normand’s statements to the police. Carol told the police that after she came home from the hospital she breast fed Rita for fifteen minutes at each breast, six or seven times a day. One of her breasts became sore, and, after calling a doctor, she purchased a breast pump. The last weekend (July 18 or 19) the *495weather got very hot, and the baby seemed to be cutting down her feeding. As a result, Carol’s breasts filled up, and she used the breast pump, and fed the pumped milk to Rita later. She got three to three and one-half ounces of milk from both breasts combined, although on one occasion she got eight ounces. Even in the last week, the baby appeared healthy to her, and seemed to be getting heavier or longer. In response to a hypothetical question, Dr. Sturner testified that if Carol had fed Rita in the manner she described the baby would not have died of starvation. For purposes of this testimony, Dr. Sturner assumed that Carol’s milk was of proper quality and quantity, and that Rita ingested milk during these feedings.
Normand told the police that he was a disabled veteran and that he stayed at home and cared for the four older children, while Carol cared for Rita. He stated that about a week before Rita died, his wife told him that the baby was not eating properly, and he told her to make an appointment with a doctor. However, he saw nothing wrong with the baby. The day before the baby died, Carol, Normand and the baby drove to a veteran’s administration hospital, where Normand was being treated. They did not seek an examination for the baby. Other than the immediate family, the only person who saw the baby after she was born and left the hospital was a babysitter. She testified that when Carol brought the baby home, it looked healthy, with “beautiful pinkish colored cheeks.” She saw Rita the Sunday before she died, and again the day before she died, and both times the baby looked paler than she had when she came home from the hospital. She never saw the baby undressed.
The defendants were charged with involuntary manslaughter. “Involuntary manslaughter is an unlawful homicide unintentionally caused by . . . wanton or reckless conduct.” Commonwealth v. Vanderpool, 367 Mass. 743, 747 (1975), and cases cited. “The essence of wanton or reckless conduct is intentional conduct, by way either of commission or of omission where there is a duty to act, *496which conduct involves a high degree of likelihood that substantial harm will result to another.” Commonwealth v. Welansky, 316 Mass. 383, 399 (1944). Commonwealth v. Gallison, 383 Mass. 659, 665 (1981). Moreover, we have approved jury instructions that “even if a particular defendant is so stupid [or] so heedless . . . that in fact he did not realize the grave danger, he cannot escape the imputation of wanton or reckless conduct in his dangerous act or omission, if an ordinary normal man under the same circumstances would have realized the gravity of the danger.” Commonwealth v. Welansky, supra at 398-399. Commonwealth v. Godin, 374 Mass. 120, 129, cert. denied, 436 U.S. 917 (1977).
Parents have a duty to provide for the care and welfare of their children. Commonwealth v. Gallison, 383 Mass. 659, 665-666 (1981), and cases cited. The Commonwealth’s position was that either Carol and Normand did not feed Rita, or they failed to take reasonable steps to care for her in light of what they knew or should have known to be her deteriorating condition, or both. As to the failure to feed Rita, based on the opinions of Drs. Mead and Sturner, the jury could have found that Rita died of starvation. However, the Commonwealth also had to produce enough evidence to permit a rational juror to infer that the baby’s condition was a result of her parents’ reckless failure to provide her with sufficient food, or reckless failure to seek medical care for the child. See Commonwealth v. Welansky, supra at 399.
Certain facts in evidence made the Commonwealth’s burden of proving recklessness particularly difficult. Just as vital to our consideration are the matters which were not shown in the evidence, and indeed were left to conjecture. There was no admission by the parents that they noticed any illness or failing of the child, and no evidence that anyone had observed deterioration of the child and had cautioned the parents. Indeed, there was no direct evidence of any kind that the parents appreciated the danger, and the case against them must necessarily arise from inferences *497drawn from the circumstances. The evidence did not support any inference of a deterioration of the child over any considerable time, because the baby died only twenty-four days after leaving the hospital, and in those twenty-four days had a weight loss of just thirteen ounces. Although the baby was dehydrated, the medical testimony was that this could occur within a few days’ time.
A particular difficulty for the Commonwealth was the proof on the issue whether the parents knew or reasonably should have known that the child was not being adequately fed. Carol stated to the police that she breast fed Rita six or seven times a day for thirty minutes, until the last week, when the baby ate less. There was no evidence that this feeding schedule was inappropriate. Dr. Sturner testified that the baby would not have died of starvation is she had been fed as Carol described. For purposes of this testimony, Dr. Sturner assumed that Carol’s milk was of proper nutritional quality, a point on which there was no evidence. He also assumed that Carol was producing milk “in sufficient quantity.” The only evidence as to quantity was Carol’s statement that when she pumped her breasts, which was not that often, and only during the last week of the baby’s life, she got three to eight ounces of milk. There was no evidence of the quantity of milk produced while breast feeding, and Dr. Sturner testified that while some mothers know whether their baby is getting milk while nursing, “some mothers don’t. I think it is a difficult situation at times. You’ve got to know the mother.” He also testified that while he assumed that the baby had “cut down” in the last week, he neither knew nor assumed how much less she ate. Earlier, he testified that even a small loss of fluid can be very crucial to dehydration, and that a person can dehydrate in a few days.
Dependent as it was on unsupported assumptions, Dr. Sturner’s testimony could not properly serve to refute the mother’s testimony as to her nursing of the child. See State Bd. of Retirement v. Contributory Retirement Appeal Bd., 342 Mass. 58, 65-66 (1961). In short (the physical ap*498pear anee of the child momentarily aside) the total evidence, including Dr. Stumer’s testimony, left to conjecture the matter of the nutritional quality of the mother’s milk and the quantity of milk ingested by the baby while nursing. See Commonwealth v. Soffron Bros., 327 Mass. 14, 16 (1951), and cases cited. Thus, there was no evidence to refute the mother’s claim that she nursed the child six or seven times a day for thirty minute periods. Disbelief of her testimony would not alone suffice to establish the opposite conclusion. See Commonwealth v. Eramo, 377 Mass. 912, 913 n.1 (1979).
It remains for us to determine whether the total evidence including the physical appearance of the child was enough to support the jury’s verdicts. The evidence as to the child’s appearance was crucial to the Commonwealth’s second theory that the defendants recklessly failed to seek medical care for Rita in light of what they knew or should have known to be her poor condition. The relevant descriptive evidence was the testimony of Drs. Mead and Sturner that Rita looked thin and malnourished, the photographs, and the babysitter’s testimony that Rita looked pale. The photographs show that the baby was woefully thin, and show the clear outline of the child’s ribs and spinal column, as well as her somewhat sunken cheeks and eyes. The testimony of the parents was that neither of them at any time noticed anything wrong with the baby other than she was eating less in the last week of her life. There was no evidence to show that Rita’s weight loss was a process of steady deterioration rather than a sudden event over a short time span. Other than the babysitter’s testimony (“pale”) and the evidence of thirteen ounces of weight loss, there was no evidence which served to compare the appearance of the baby when she came home from the hospital and her appearance at the time of her death. Nor was there any evidence on the issue whether the child’s appearance, as shown by the photographs taken several hours after death, was fairly representative of her appearance before death.
*499We think the total evidence, including that relating to the child’s physical appearance, was not sufficient to establish reckless culpability. A case could be made that the defendants were negligent as measured by the standard of reasonable parents, but recklessness is more than a mistake of judgment or even gross negligence. Commonwealth v. Godin, 374 Mass. 120, 127 (1977). Commonwealth v. Welansky, 316 Mass. 383, 400-401 (1944). We have stated that it is “conduct [which] involves a high degree of likelihood that substantial harm will result to another,” Commonwealth v. Welansky, supra at 399, or “which constitutes ... a disregard of probable harmful consequences to another,” Commonwealth v. Vanderpool, 367 Mass. 743, 747 (1975). The evidence here fell markedly short of that produced in Commonwealth v. Gallison, 383 Mass. 659, 665-666 (1981), quoting from Commonwealth v. Godin, 374 Mass. 120, 130 (1977) (“The evidence was susceptible of the finding that the mother’s inaction in light of her child’s vomiting, diarrhea, high fever, subsequent unconsciousness, and breathing failure created a ‘substantial and unjustifiable risk’ of death”). We conclude that the Commonwealth failed to produce evidence sufficient to permit a rational juror to find beyond a reasonable doubt that the defendants’ actions were reckless. Therefore, the judgments of conviction are reversed, and verdicts are set aside, and the cases are remanded to the Superior Court for the entry of required findings of not guilty.