SUMMARYA theoretical comparison is made between two different criteria of statistical significance for detecting abnormalities in a scan. Criterion I utilizes as much of the available information content as possible, while Criterion II is applicable to visual reading of a photoscan or colour‐scan. The limits of detecting abnormalities with Criterion II is theoretically inferior to that with Criterion I.Using theoretical estimates of contrast for various sized abnormalities, the concentration ratio of 99mTcO4 in cerebral lesions is estimated from a number of scans. The mean concentration ratio was 6‐5, assuming a log‐normal distribution.