Cancer survival in Sweden in 1961-1991 is presented as a comprehensive report from the Swedish Cancer Registry. The report shows both successes and failures, confirms some earlier published results and presents some new findings worth further analysis. Survival has increased for female breast cancer, malignant melanoma, cancers of the testis and thyroid gland, acute leukemia, and Hodgkin's disease. No improvements are found for multiple myeloma or cancers of the liver, gall bladder, and pancreas. Small increases are shown for colorectal cancer and cancers of the stomach, oesophagus, and kidney. Increases in postoperative survival are shown for sites dominated by histologically benign tumors, i.e., intracranial neurinoma, meningioma, and cancers of the endocrine glands such as parathyroid tumors. From 1970-1972 to 1980-1982 the 10-year relative survival rate (RSR) increased from 30% to 38% for males and from 44% to 51% for females. Hence, cancer survival for all cases combined has approached the survival of the general population somewhat. Most of the increases took place in the 1970's. Changes in the distribution of incidence towards cancer sites with better prognoses account for some 10-20% of the observed increases in RSR, whereas the aging of the cancer population reduces the upward trend in RSR for all cases combined by some 1-2%. Cancer patients have poorer survival than the population long after 5 years of follow-up. They reach the survival of the population after about 8-12 years for colorectal cancer, 10 years for cervical cancer, 7-10 years for malignant melanoma, 13-18 years for kidney cancer, and more than 19 years for female breast and prostate cancer. For patients diagnosed in 1970-1972 this occurred 16 years after diagnosis at 29% for males and 43% for females when all cancer cases were combined. The extended time until 'statistical cure' for most cancer forms clearly indicates the need to augment the commonly used 5-year RSR with other outcome measures. If cancers on average are discovered earlier today, the 5-year RSR gives an exaggerated impression of the improvement over time. In this case the change in the 10-year RSR is a less biased criterion.