A total of 3784 women aged 35 and above living in 26 villages in rural Maharashtra state, India, were invited to undergo a pelvic examination, to evaluate the performance of unaided visual inspection by trained paramedical workers in detecting cervical cancer. Of this number, 2135 (56.4%) women complied with the invitation. Paramedical workers scored 1120 (57.3%) and 118 (6%) women as having abnormal cervices using the low- and high-threshold criteria respectively. There was good agreement between the visual findings of the paramedical workers and those of a gynaecologist. All subjects had a cervical smear. A total of 10 cervical cancers were detected by cytology/histology. The sensitivity of visual inspection by paramedical workers to detect cervical cancer was 90.0% using the low threshold and 60.0% with the high threshold to define a positive test. The values for specificity were 42.8% and 94.5% respectively. The results obtained by the gynaecologist were very similar. Cost savings implied by limiting cytology/other investigations to approximately half of the population pre-selected on the basis of visual inspection are likely to be offset by the necessity to repeat the test at frequent intervals, repeated follow-up visits and other investigations.