When the billing of local telephone service is changed from flat rate to measured service, the distribution of monthly calling rates is altered. This report models the distribution of flat-rate telephone usage in terms of demographic variables and stochastic components; the shift to measured service affects both the systematic and stochastic parameters. The model is fitted by maximum likelihood to data for interviewed households participating in General Telephone's local measured service experiment in Illinois. Households tend to make more calls if they are larger (more people), older, or include teenagers. They tend to reduce calling proportionately more in response to usage charges if they average many calls under flat rate for any of the above reasons or for other, unexplained reasons. There is substantial variation in telephone usage by households with similar demographic characteristics. Consequently, the benefits and costs of local measured service will tend to be diffused across demographic groups.