Variability in patient positioning was determined by analyzing simulation and portal film measurements for 318 portals in 51 patients treated with external beam radiotherapy to the head and neck. Several indicators of error in patient positioning were examined: random error, a measure of the deviation of all portal films from the average portal film position, systematic error, a measure of the difference between the average portal film and the simulation film, and total uncertainty, a measure of the overall deviation, including both random and systematic uncertainties. The median differences noted were 0.4 cm, 0.6 cm, and 0.7 cm, for Random Error, Systematic Error, and Total Uncertainty, respectively. The treatment fields analyzed in this study show a substantial treatment-to-treatment and simulation-to-treatment variability in patient positioning. The methods described provide an improved means for the systematic analysis of variability in patient positioning.