The efficacy of Papanicolaou smear screening intervals of longer than 1 year is uncertain. Ninety-two symptomatic cases of invasive cervical cancer diagnosed between 1978-1983 in western Washington were identified with a population-based tumor registry. Using a random-digit-dialing technique, 178 controls from the same geographic area were selected. A structured telephone interview was used to ascertain screening history and risk factors for cervical cancer. The mean interval between papanicolaou smears in the 10 years preceding diagnosis (cases) or the reference date (controls) was calculated. The risk of squamous cell cervical cancer was increased 3.9 times (95% confidence interval 1.2-12.3) for women with Papanicolaou smears at 3-year intervals compared with women with annual screening. For women who had not had a Papanicolaou smear in the preceding 10 years, this risk increased 12.3 times (95% confidence interval 2.5-60.6). For screening intervals of 2 years, the risk of cervical cancer (all cell types) was not increased (relative risk 1.01; 95% confidence interval 0.43-2.37). The presence of well-known risk factors for cervical cancer did not modify these results. These data suggest an increasing risk of cervical cancer if Papanicolaou smear screening intervals exceed 2 years.