Cervical smears were reviewed from patients in whom a cytological abnormality was followed, after an interval without interference, either by regression to ;negative' or else by progression to invasive carcinoma. Twenty-eight cases were from a previously analysed series with positive smears and an interval of at least two years before investigation, resulting from refusal or failure to trace. Slides were also reviewed from 25 cases in which ;positive' smears had regressed to negative without escaping from surveillance, and from 10 patients subsequently developing invasive carcinoma whose previous slides, taken several years earlier, showed abnormalities on review. None of these 63 patients had any biopsy or other surgical procedure to the cervix between the initial smear and the outcome. Slides showing ;superficial cell dyskaryosis' and/or well-differentiated ;parabasal cell dyskaryosis' were found only among the groups with subsequent regression. Those showing dissociated poorly differentiated dyskaryotic parabasal cells regressed to negative in two cases and progressed to invasion in nine. This suggests that many examples of spontaneous regression correspond to mild dysplasias which are not precancerous, and overdiagnosis must often have resulted in unnecessary surgical procedures in the past.;Regressing' and ;progressing' groups both included cases in which the spatula had removed coherent pieces of undifferentiated epithelium. These are difficult to interpret cytologically. In nine of them (including four which regressed) the cytological picture was that of carcinoma in situ. The remainder (14 cases) were probably examples of reserve cell hyperplasia, and it is noteworthy that, of the 21 cases subsequently progressing to invasive carcinoma, five were preceded by appearances of this type. It is concluded that cell aggregates suggesting an unusual degree of reserve cell hyperplasia are a danger signal and require careful surveillance.