Comparisons of outcomes of treated and untreated patients in observational studies may be biased due to differences in patient prognosis, often as a result of unobserved biases in patient selection for treatment. We analyzed a cohort of 122,124 heart attack patients, linking chart data with health administrative data. Invasive vs. conservative cardiac management was associated with a 50% relative decrease in mortality using standard multivariate modeling, propensity score risk adjustment, or propensity-based matching. Instrumental variable analyses showed a 16% relative decrease in mortality, using regional cardiac catheterization rate as the instrument. The sensitivity to analytic method is primarily due to survival bias since short term mortality rates are high. Caution is advised in analyzing the effects of invasive treatment using standard statistical methods with observational data.