The records of all 1591 women with a histologically confirmed primary breast neoplasm who received their primary treatment at the main referral centre in British Columbia and were diagnosed in the years 1945, 1950, 1955, 1960, 1965, 1970, or 1975 were reviewed. The interval from appearance of the first symptom to diagnosis decreased from 1945 to 1960, but no change was seen from 1960 to 1975. An analysis of survival from the date of first symptom showed that long-term survival was greater in patients with a shorter delay between the appearance of symptoms and diagnosis. The demonstration that shorter delay does improve survival, even when assessed from the appearance of the first symptom, yet delay times have not been falling recently, suggests that educational efforts are inefficient.