The dose of radioiodine (131I) used to survey patients for metastatic functioning thyroid cancer varies from 0.2 mCi to 30.0 mCi. Higher doses have occasionally revealed more tumors, but deliver more radiation to the patient. We asked which dose would be sufficient to detect metastatic deposits. Using a water tank with small-source phantoms, we sought to determine: the minimum volume and concentration of activity capable of being imaged, effects of background and source depth on detectability, and a practical 131I tracer dose based on these findings. Two gamma cameras affixed with high-energy collimators of different design were used to evaluate the capabilities of two instrument systems. The lowest activity detectable at the water surface was 0.03 microCi, in volumes of 10 to 300 microliters. Background activity at 0.01 microCi/ml resulted in a three to tenfold loss of detectability; computer subtraction of background did not improve results. We assumed that the minimum beneficial treatment would be 4,500 rad, a dose delivered by 200 mCi of 131I to a tumor with 0.05% uptake of the dose per gram. From these assumptions, our data show that a 2 mCi diagnostic dose would detect 10 and 30 microliter lesions containing 0.05% or more of the dose per per gram, but only at the surface and in the absence of background radioactivity. Moreover, assuming patient motion and background activity, some potentially treatable lesions probably cannot be detected even with a 30 mCi diagnostic dose, using present-day equipment. Selection of a diagnostic dose should therefore acknowledge the limitations of scintigraphic detection and take into account the radiation burden incurred by studies repeated over years.