The Zimbabwe national cancer registry was established in 1985 as a population-based cancer registry covering Harare city. Cancer is not a notifiable disease, and registration of cases is done by active methods. The registry contributed data on randomly drawn sub-samples of Harare resident cases among 17 common cancer sites or types registered during 1993-1997 from black and white populations. Follow-up was carried out predominantly by active methods with median follow-up ranging from 1-54 months for different cancers. The proportion with histologically verified diagnosis for various cancers ranged from 20-100%; death certificate only (DCO) cases comprised 0-34%; 58-97% of total registered cases were included for survival analysis. Complete follow-up at five years ranged from 94-100%. Five-year age-standardized relative survival rates of selected cancers among both races combined were cervix (42%), breast (68%), Kaposi sarcoma (4%), liver (3%), oesophagus (12%), stomach (20%) and lung (14%). Survival was markedly higher among white than black populations for most cancers with adequate cases. Five-year relative survival by age group was fluctuating, with no definite pattern or trend.