The study objectives were: 1) to analyse the incidence and death rates from cancer among children aged 0-14 years resident in the north-eastern Italian province of Trieste between 1972-1993, using data from the population-based Trieste Cancer Registry; 2) to evaluate the local diagnostic facilities by analysing the accuracy of histological diagnoses, the causes of delay in the diagnosis, and the interval between onset of symptoms and diagnosis of cancer; 3) to calculate the proportion of patients treated following the most effective therapy protocols known at the time of the tumour detection, and to compute the actuarial five-year survival rates since diagnosis. We recorded 123 new cases of cancer (93% microscopically verified) corresponding to a rate, age-standardized to the world population, of 161.9 (standard error [SE] = 15.1) per million child-years. The most common diagnostic group was that of primary brain tumours: 40 cases, rate = 51.0 (SE = 8.4). In 102 cases the diagnosis was made at hospitals in the province of Trieste, with a median time of seven days (25th-75th percentile = 1-16) between admission and diagnosis. In 37 cases the length of the interval between the advancing of the diagnostic hypothesis of cancer and the microscopic diagnosis ranged from eight to 57 days: 20% of the interval was spent in the identification of the lesion, 50% elapsed between the identification and the biopsy, and 30% was spent in performing the microscopic diagnosis. Out of 123 cases, 30 were partly treated or completely treated at centres not located in our province, i.e., at seven different Italian hospitals (14 cases), nine European hospitals (15 cases), and at one North-American centre (one case). The 40 children with brain tumours were spread among 12 institutions. The five-year survival rate increased from 52.4% (SE = 6.3) for the 63 children with cancers diagnosed in 1972-1981 to 62.5% (SE = 7.0) for the 48 with malignancies detected in 1982-1990.