The impact of insurance on access to physician services among elderly individuals with chronic illnesses has far-reaching policy implications. Using a national probability sample of aged Medicare beneficiaries (N = 5,543), and controlling for severity of illness, comorbidities, and other covariates, we analyze this issue for the most prevalent, chronic disabling disease among the elderly: arthritis. The results from the two-part multivariate model (logistic regression followed by ordinary least squares regression) suggest that insurance status is a positive and statistically significant predictor of both initial access to care (p < .01), and amount of arthritis care used (p < .05). Taking the results of the two-part model, we conducted a microsimulation to estimate the increase in Medicare spending that would result if various types of Medicare supplemental insurance were provided to those who had none. Results suggest that Medicare expenditures would rise from $51 million to $59 million annually depending on the type of supplement provided.