A significant effect of treatment duration on pelvic control was found in 830 patients with cervix cancer treated by radical radiation therapy. Using three methods of analysis, the loss of control consistently approximated 1% per day of treatment prolongation beyond 30 days, although analysis of stage subgroups showed that this effect was predominantly manifested in Stages III/IV compared with Stages I/II. In multivariate analyses using both a logistic regression and a Cox regression model, stage (p = 0.0001 for Stage I/IIA and 0.0036 for Stage IIB relative to Stage III/IV) treatment time (p = 0.0001), and age (p = 0.0067) were independently correlated with pelvic control. Exclusion from analysis of patients with delays due to tumour or treatment related complications, intercurrent illness or manifestations of poor tumour response did not significantly change the magnitude of the time effect nor the ranking of the significant covariates. These results are consistent with the occurrence of accelerated repopulation and warrant further investigation, preferably in a randomized trial of accelerated versus conventionally fractionated radiation therapy.