Studying worker health insurance choices is usually limited by the absence of price data for workers who decline their employer's offer. This paper uses a new Medical Expenditure Panel Survey file which links household and employer survey respondents, supplying data for both employer insurance takers and declines. We test for whether out-of-pocket or total premium better explains worker behavior, estimate price elasticities with observed prices and with imputed prices, and test for worker sorting among jobs with and without health insurance. We find that out-of-pocket price dominates, that there is some upward bias from estimating elasticities with imputed premiums rather than observed premiums, and that workers do sort among jobs but this does not affect elasticity estimates appreciably. Like earlier studies with less representative worker samples, we find worker price elasticity of demand to be quite low. This suggests that any premium subsidies must be large to elicit much change in worker take-up behavior.