The influence of age on Group Health medical claims is analyzed for a large data set provided by the Society of Actuaries. This data set comprises claims for the years 1997 through 1999, with a total claim count of about 4.3 million and total paid charges of approximately $7 billion. Using partial linear models that allow for the influence of age to be nonlinear, it is shown how age affects claim severities and claim relativities (defined as the proportions of claimants by diagnostic category). The claim severity model uses quantile smoothing with total variation regularization in the nonparametric component. The claim relativities approach rests on a multinomial logit model with a random-walk smoother. The findings show that the influence of age on claim severities is log-linear within an age bracket that ranges from the early twenties to the early sixties; the effect of age on claim relativities is highly nonlinear.