The purpose of this study is to compare two ways of estimating the ratio of internal noise variability to external noise variability (σ1/σE). One method [D. M. Green, Psychol. Rev. 71, 1392–1407 (1964)] infers internal noise from the consistency a listener shows in choosing between two different noise maskers when the same pair are repeated on another trial. The second method [R. A. Siegel, masters thesis, M.I.T. (January 1979)] infers internal noise from the difference in signal detectability between trials containing a pair of identical versus different maskers. The masker bandwidth (narrow = 100 Hz or wide = 2800 Hz) and mode of presentation (continuous, burst, or continuous with gaps surrounding the observation intervals) were varied. The observation interval was 10 ms. Both methods estimate the ratio (σ1/σE) to be within experimental error of unity for most conditions. Factors that influence the estimate will be discussed, such as temporal uncertainty, response bias, binomial variability in the estimate of detection performance, and incoherence between signal and masker. [Work supported in part by a grant from the National Institutes of Health, Public Health Service, U.S. Department of Health, Education and Welfare.]