Testicular cancer has a peak incidence at around 30 years of age, and it is the most common cancer in men under 35 years of age, accounting for a third of all cancers in that age range (Adami et al, 1994). The age distribution is unusual among adult malignancies, with almost 70% of cases arising in men under 40 years of age, and only about 2% in men over 70 years of age. Incidence has increased steadily in many developed countries over the last 50 years, including the United Kingdom (Coleman et al, 1993; Richiardi et al, 2004). In men under 50 years of age, incidence has risen almost two-fold in England and Wales in the last 30 years, but the increase has been much smaller among older men (Quinn et al, 2001). Age-standardised mortality was 1 per 100 000 per year during the 1950s, and it rose slightly during the 1960s and 1970s. Advances in the treatment of testicular cancer since the 1960s have had a dramatic effect on survival (Coleman et al, 1999), and men born during the 1950s and 1960s had substantially lower testicular cancer mortality in the late 1970s and early 1980s than earlier birth cohorts, despite the substantial rise in incidence (Quinn et al, 2001). Cryptorchidism is one of the few known risk factors (Richiardi et al, 2002). A positive family history also carries a higher risk, but vasectomy has not been shown to increase risk (Moller et al, 1994). Most testicular cancers are germ cell tumours, either seminoma or non-seminoma. Seminomas arise from the sperm-producing testicular germ cells, whereas non-seminoma germ cell tumours include teratoma, choriocarcinoma and yolk sac tumours. Seminomas are treated by surgical removal of the affected testicle (orchiectomy) and adjuvant radiotherapy. Surgical treatment is also effective with early stage teratomas. Orchiectomy alone cures around 80% of testicular cancer patients (Dearnaley et al, 2006). We analysed survival patterns for 18 605 men diagnosed with testicular cancer aged 15–99 years in England and Wales during 1986–1999 and followed up to 31 December 2001. Less than 5% of tumour records had to be excluded from analysis: 2% because the vital status was not known when data were extracted for analysis, less than 2% because the testicular cancer either followed or was synchronous with another primary malignancy, and about 1% because the date of diagnosis was the same as the date of death (‘zero survival'). Virtually all the tumours in the last category will have been registered solely from a death certificate, from which the date of diagnosis is unavailable, as very few men would be expected to have died on the day of diagnosis, but the two categories could not be reliably distinguished in the national data. In total, 54% of tumours were seminomas, and 29% teratomas, similar to the proportions seen in the late 1980s (50 and 30%). Rare morphologic types included choriocarcinoma, trophoblastic teratoma (1%), Leydig cell tumour (0.3%) and a few Sertoli cell tumours. Although cryptorchidism is a known risk factor, less than 1% of tumours were recorded as having arisen in an undescended testis.