This study investigated the possibility of improved survival in cerebral palsy (CP) over a 20-year period. Participants were 47 259 persons with CP receiving services from the State of California between 1983 and 2002. The person-year approach was used. This asks whether the probability of dying in a given calendar year changes over the study period after age and severity of disability are taken into account. An appreciable improvement over time was found in children with severe disabilities and in adults who required gastrostomy feeding. In these groups, mortality rates fell by 3.4% per year. Therefore, life expectancies reported in earlier studies should be increased by approximately 5 years if adjustments to 2002 mortality rates are made. For other persons with CP there was, at most, a small improvement over the 20-year period. The results suggest there have been improvements in the treatment and care of the most medically fragile children. Gastrostomy feeding has become much more widespread over the past two decades, and the improved survival of persons with gastrostomies may reflect better understanding of their requirements.There is little doubt that survival in cerebral palsy (CP) has improved over the past half-century. 1 However, no clear improvement in survival over the past two to three decades has been documented in the literature on CP. Although Strauss et al., 2 in their Californian study of children with CP, observed some improvement over time, it was not statistically significant. No evidence for such a trend was found in the Australian study of Blair et al., 3 the UK studies of Hutton et al, 4,5 and the Californian study of adults by Strauss and Shavelle. 6 Such an improvement would be an example of a secular trend. Secular trends have been reported in spinal cord injury over a 20-year period, 7,8 and for infants in a vegetative state over a 15-year period. 9 Secular trends in survival are of interest for at least two reasons. First, when referring to longevity studies in the literature the reader will want to know whether the published estimates need to be revised upwards and, if so, by how much. Second, such trends may be a marker for progress in medical treatment and therapy. For example, a 1992 article on fundoplication in children with profound neurological disability (Smith et al. 10 ) was subtitled 'High risks and unmet goals'. However, anecdotal evidence suggests that since the 1980s there has been a better understanding of the importance of appropriate nutritional status in children and adults with disabilities. Further, there has been a greater appreciation of the risks of aspiration, improvements in surgical procedures for gastrostomy and fundoplication, and progress in gastrostomy-feeding techniques. As a result, it seems unlikely that Smith's subtitle would apply today. Nevertheless, so far, no such improvement appears to have been documented in a published study on mortality or morbidity in CP.To identify a possible secular trend in survival in CP, it is necessary to control for fa...