Rabbit (1968) tested the accuracy with which young adults with excellent hearing could repeat lists of numbers, words or continuous text which they heard over a system which maintained a constant ratio of white noise to speech signal irrespective of signal strength. Noise levels which allowed individual subjects correctly to repeat every word as they heard it (i.e. shadow without errors) nevertheless reduced their ability to remember what they had heard. This appeared to be because increased effort necessary to recognise words through low levels of noise prevented adequate rehearsal or elaborative encoding of material to be remembered. This finding that slight degradation of sensory input had secondary consequences on memory and comprehension of spoken material led to an interpretation of findings that 960 individuals aged from 50 to 82 years, in contrast to young adults, showed markedly better recall for word lists presented visually than for word lists presented auditorally, even when each word in each list was correctly read or repeated aloud. Audiometric screening of a sub-set of this population allowed 30 individuals with mild hearing losses (35 to 50 db) in each of the three age-decades 50 to 59, 60 to 69 and 70 to 79 years to be compared with matched controls who had pure-tone hearing losses of less than 35 db on memory for lists of 30 words presented visually or over a good-quality sound system. In both presentation modes lists were only scored if all words were correctly read or repeated. Individuals with slight hearing loss recalled visually presented words as well, but recalled auditorally presented words significantly less well, than their controls.(ABSTRACT TRUNCATED AT 250 WORDS)