This study was conducted to validate the night vision threshold test (NVTT) as an indicator of night blindness. A total of 1401 pregnant women from the National Maternity Hospital participated in this study. Women were queried about night blindness and took the NVTT using standardized procedures after 10 min of dark adaptation. Sixteen percent failed the NVTT, but only 6.4% reported having night blindness. Blood samples from women who failed the NVTT (cases) and matched controls indicated the serum vitamin A (SVA) concentration was lower (P < 0.05) in cases (1.19 +/- 0.03 micromol/L) than in controls (1.29 +/- 0.03 micromol/L). The SVA concentrations did not differ between women who reported and did not report night blindness. The SVA concentration was correlated (r = 0.22, P < 0.001) with the NVTT scores. Twenty-five percent of women with an SVA < 0.35 micromol/L reported night blindness while 100% failed the NVTT. Nineteen percent of women with an SVA < 0.70 micromol/L reported night blindness while 73% failed the NVTT. A receiver operating characteristics analysis indicated that the NVTT had greater sensitivity (0.73 vs. 0.19) and less specificity (0.51 vs. 0.87) compared with reported night blindness for women with SVA < 0.70 micromol/L and greater sensitivity (100.0 vs. 0.73) and similar specificity (0.51 vs. 0.50) for women with SVA < 0.35 micromol/L. The NVTT identified women with low SVA and self-reported night blindness was misleading. We provide a preliminary algorithm to predict the population of women with low SVA concentrations.