OBJECTIVE To investigate the contrast thresholds (CTs) in normal subjects using a high-luminance, letter-recognition task under clinically relevant testing conditions. METHODS Texas Tech University Health Sciences System, Lubbock, Texas, USA. METHODS Sixty normal subjects aged 20 to 49 years with a best corrected visual acuity of 20/20 or better in both eyes participated. M & S Technologies software was used to display black-on-white Sloan letters at contrast levels of 25%, 20%, 15%, 12%, and 10% through 1% in 1% decrements. The effects of age, sex, optotype size, eye dominance, ambient illumination level (bright = 625 - 630 lux; dim = <3 lux), and direction of approach to threshold were analyzed using a multivariate, ordinary, least-squares analysis. RESULTS Age and sex did not influence CTs. Ascending versus descending testing was not statistically significant (P>.5). The effects of room illumination and eye dominance were significant (P<.01). Significant differences were found between 20/30 and 20/50, 20/30 and 20/70, and 20/50 and 20/70 optotype sizes (P<.01 for all comparisons). CONCLUSIONS A commercially available, computer-based test of CTs was easy to administer and apparently easy for inexperienced subjects to perform. The results suggest criteria for detecting visual problems concerned with familiar but complex spatial-image shapes. This information might be used to assess the effects of treatments such as laser refractive surgery on recognition contrast. Further study is warranted.