Both diazoxide and diphenylhydantoin have been shown to cause hyperglycemie in man and to inhibit insulin secretion in vitro. The effects of these two drugs upon the response to 300 mg./100 ml. glucose were contrasted in the isolated, perfused rat pancreas. Similarities: Both drugs inhibited within seconds. At high concentrations (75 μg./ml.) of either drug, 95 to 100 per cent inhibition occurred. Upon withdrawal, return of secretion was rapid. Differences: During constant (five- and twenty-minute) infusions of diazoxide (10 to 75 μg./ml.), there was an initial fall in secretion and then an “escape” toward pre-inhibition levels; after diazoxide there was a postinhibitory overshoot. In contrast, five-minute infusions of diphenylhydantoin (5 to 75 μg./ml.), and twenty-minute infusions of 25 to 75 μg./ml., though causing comparable levels of initial inhibition, did not cause escape or postinhibitory overshoot. However, an incompletely inhibitory concentration of diphenylhydantoin (10 μg./ml.), given for twenty minutes, was followed by this overshoot. Computer simulation, based on the compartmental-quantal model, suggested that diphenylhydantoin inhibits both secretion from a labile insulin compartment and provision of insulin and/or precursor to this compartment. Lower, partially inhibitory concentrations (i.e. 10 μg./ml.) of diphenylhydantoin might allow for some provision to proceed, and thus a postinhibitory overshoot would occur. The escape and overshoot noted with all inhibitory levels of diazoxide, when compared with the simulations, suggested that this drug may act primarily to inhibit a late step in the provisionary phase.