This case involves a three-way struggle over the use of Compound 1080, a preda-cide used to kill coyotes and other predators threatening the economic well-being of the livestock industry. Caught in the middle of the struggle is the EPA, which decided to lift the ban on Compound 1080, but in doing so imposed severe restrictions on its use. Defenders of Wildlife argue that all uses of Compound 1080 should remain banned. The National Cattlemen’s Association, on the other hand, not only applauds the EPA’s decision to lift the total ban on Compound 1080 but would like to see the restrictions placed on its use removed.
Prior to 1972, Compound 1080 was one of the principal substances used by the United States government to control predation on livestock. In 1972 the EPA determined that the use of Compound 1080 was leading to substantial deaths of non-target wildlife — wildlife which was not predatory and which was being either directly poisoned by feeding on the baits or secondarily poisoned by-feeding on the carcasses of target animals which had been poisoned. The EPA therefore cancelled the registration of Compound 1080 for use as a predacide pursuant to its authority under the Federal Insecticide, Fungicide, and Rodenticide Act (F.I.F.R.A.). 7 U.S.C. § 136d(b) (1982).
An applicant seeking to re-register a can-celled predacide must demonstrate the existence of “substantial new evidence which may materially affect the prior cancellation or suspension order and which was not available to the Administrator at the time he made his final cancellation or suspension determination.” 40 C.F.R. § 164.131(a) (1984).- The EPA will then hold a hearing to determine whether the substantial new evidence “requires reversal of modification of the existing cancellation or suspension order.” 40 C.F.R. § 164.132(a) (1984).
In 1982 the EPA ordered a hearing to evaluate evidence that may have become available since 1972 to determine whether “substantial new evidence” mandated lifting the ban on Compound 1080. After the hearing,' the ALT determined that, while the ban on some uses of Compound 1080 should continue, substantial new evidence did exist that certain uses of Compound *2701080 could be allowed without serious risk to the environment.
Three methods of Compound 1080 delivery were considered by the EPA and form the areas of dispute in this case. Before 1972, the predominant method of using Compound 1080 was in what are known as large bait stations. A large bait station consists of a fifty to one-hundred-pound portion of horse or sheep carcass impregnated with Compound 1080. These bait stations were set out during the winter and early spring in rangelands suffering from heavy predation by coyotes. The AU determined that the proponents had not met their burden of showing substantial new evidence justifying reversal of the ban on use of 1080 in large bait stations and concluded that the ban on that use should continue.
Two new methods of Compound 1080 delivery have been developed since 1972. The first method involves using what are known as single lethal dose baits or SLDs. An SLD consists of a bite-size piece of meat or other material containing a lethal dose of Compound 1080 which is placed in a location where it is likely to be taken by a coyote and not likely to be consumed by non-target wildlife. The second method involves use of toxic collars, which consist of rubber collars with small reservoirs filled with Compound 1080 solution which are worn by sheep or goats. This method utilizes the fact that when a coyote preys on livestock, it generally strikes first at the animal’s throat. Theoretically, a coyote which attacked an animal wearing a toxic collar thus would likely pierce the Compound 1080 reservoirs and be exposed to a lethal dose of the substance. The AU determined that all of the evidence with respect to the use of SLDs and toxic collars had to be considered new evidence since these delivery methods were not in use at the time of the original ban in 1972. Based upon the evidence, the AU determined that SLDs and toxic collars, when used as prescribed, did not pose a serious threat to the environment.
In order to prevent unauthorized uses, the AU issued use restrictions for both toxic collars and SLD baits. Under the AU’s order, certification for all 1080 uses would be supervised by a federal agency. Collars could be filled and distributed only by registered users and could be used only by certified applicators or persons under their direct supervision. There would be no restriction on who could become certified to use collars, and it is anticipated that ranchers themselves could do so if they were qualified. SLD baits, on the other hand, could be prepared, distributed and used only by federal and state employees who had been certified. The AU also imposed other restrictions on the placement of SLDs, and required recordkeeping with respect to their effectiveness and hazards.
All of the parties appealed the AU’s initial decision to the EPA Administrator. The Administrator affirmed the AU’s decision with respect to large bait stations, noting “the hazards created by the Compound 1080 large bait station, and the lack of evidence as to its efficacy.” Final Decision at 23. He also affirmed the decision to lift the ban on use of 1080 in single lethal dose baits, but imposed additional restrictions on that use, including a requirement that the certification process be run solely by the federal government. In addition, he stated that additional testing under experimental use permits would be required before actual registration would be allowed. Final Decision at 25. Finally, he affirmed the decision to lift the ban on use of Compound 1080 in toxic collars. He imposed additional usage and labelling restrictions on the use of 1080 in this form, however, and stated that final use restrictions should be developed in the subsequent registration proceeding. Final Decision at-31.
Defenders of Wildlife appeal, asserting that the Administrator’s decision to lift the total ban on Compound 1080 is not supported by substantial new evidence. On the other hand, the National Cattlemen’s Association, et al appeal, claiming that the EPA’s decision to continue its ban on the use of large bait stations was improper and *271that certain of the EPA’s restrictions on the use of SLD baits are neither supported by the record nor authorized by law. The National Cattlemens Association also complains about the EPA’s refusal to consider other bait delivery methods which were determined by the ALJ to be outside the scope of the hearing.
The Administrator’s order must be sustained “if it is supported by substantial new evidence when considered on the record as a whole.” 7 U.S.C. § 136n(b) (1982). Substantial evidence “is more than a scintilla. It means such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Consolidated Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 217, 83 L.Ed. 126 (1938). However, the possibility of drawing two inconsistent conclusions from the evidence does not prevent an administrative agency’s findings from being supported by substantial evidence. Consolo v. Federal Maritime Commission, 383 U.S. 607, 620, 86 S.Ct. 1018, 1026, 16 L.Ed.2d 131 (1965); Environmental Defense Fund, Inc. v. EPA, 548 F.2d 998, 1003 (D.C.Cir.1976), cert. denied, 431 U.S. 925, 97 S.Ct. 2199, 53 L.Ed.2d 239 (1977). In addition, “when questions involve a special expertise of an agency, such as in detailed scientific proceedings, the agency deserves special deference from the courts, although careful review is always required.” Environmental Defense Fund, 548 F.2d at 1252.
We turn first to the. claims of the Defenders of Wildlife. We find that the EPA’s decision to lift the ban on Compound 1080 for use in the newly discovered delivery methods is supported by substantial new evidence as that standard has been defined. The EPA’s decision to lift the ban on Compound 1080 use in SLD baits and allow experimental use permits for those baits rests mainly on the successful results from the use of SLD baits in British Columbia during 1982-83. The British Columbia program supports the EPA’s conclusions that SLD baits can be effective in controlling local predation problems and that they have little impact on non-target animals when properly placed and carefully monitored. In this evolving field of continuing study and controversy, we believe this study is “substantial” evidence in support of the EPA’s decision when viewed in light of the record as a whole. While there was other evidence in the record tending to show that SLD baits are ineffective and pose risks to non-target wildlife, there is sufficient new evidence to validate the EPA’s conclusion that carefully controlled SLD baits could be both effective and safe.
Likewise, an overview of the record shows that there is substantial new evidence to support the EPA’s determination that toxic collars will be both effective and safe. Virtually all this evidence has been developed since the 1972 imposition of the ban and therefore must be considered “new.” While there is some evidence that toxic collars could pose a danger to the environment if abused or negligently handled, it was not unreasonable for the EPA to conclude that these risks are so slight that they should not prevent use of properly regulated collars. Thus we conclude that the EPA was correct in finding that there was substantial new evidence to support lifting the ban on Compound 1080 for use in toxic collars. By the same token, while reasonable minds could disagree with their decision, there is plainly substantial evidence in the record as a whole to support the EPA’s conclusion that the proponents of large bait stations had not introduced substantial new evidence which would justify reversing or modifying the ban on use of Compound 1080 in that delivery system.
Turning to the claims of the National Cattlemen’s Association, we find that the concerns raised by Defenders of Wildlife with respect to the hazards posed by the use of Compound 1080 in SLDs and toxic collars should these delivery methods be abused or improperly administered justify the restrictions • imposed by the EPA on these uses. While the National Cattlemen’s Association objects to most of the restrictions placed on these uses, their principal concern appears to be the EPA’s re*272striction preventing anyone but authorized state and federal employees from administering SLDs. The Association asserts that it is inconsistent to allow ranchers to personally supervise the use of toxic collars but not the use of SLDs. We find, however, that there is no inconsistency in the EPA’s decision. Toxic collars require no preparation by the farmer or rancher but need merely be placed around the neck of the target animals. In addition, evidence introduced at the hearing, supports the conclusion that, because of the cost of the toxic collars, it is highly unlikely that the collars will be purchased for the purpose of extracting the Compound 1080 for use in unauthorized delivery methods. On the other hand, it would be extremely difficult to prevent the unauthorized use of 1080 in illegal delivery methods should private individuals be licensed to prepare and administer the SLDs. It is clear that the EPA’s decision is subject to significant dispute. However, we must affirm if that decision is supported by substantial new evidence. We must conclude that it was reasonable, in light of the record as a whole, for the EPA to limit the use of SLDs to properly authorized state and federal employees.1
Finally, the Association argues that the EPA exceeded its authority when it required that a federal agency determine the competency of and certify all users of SLD baits. In support of its objection, it points to section 4(a)(2) of the Federal Insecticide, Fungicide and Rodenticide Act, 7 U.S.C. § 136b(a)(2) (1982).2 The language of this subsection supports the Association’s position. In addition, section 136c(f), which governs experimental use permits, provides:
Notwithstanding the foregoing provisions of this section, the Administrator shall, under such terms and conditions as he may by regulations prescribe, authorize any State to issue an experimental use permit for a pesticide. All provisions of section 136b of this title relating to State plans shall apply with equal force to a State plan for the issuance of experimental use permits under this section.
Thus, the Administrator is without statutory authority to adopt a blanket rejection of state plans for certification, notwithstanding the evidence of widespread abuse by state plans documented by the Administrator. The Administrator may reject individual state plans if those plans do not provide adequate assurance of compliance with the EPA’s requirements. Such rejection cannot, however, be arbitrary and capricious and cannot be made in advance of application by the state.
We find that the remainder of the EPA’s restrictions on the use of toxic collars and *273SLDs are within the standards set by the law and that the need for the restrictions is supported by substantial evidence in the record.
The Administrator’s blanket rejection of state plans is reversed. The final decision is in all other respects affirmed.