Mass differences less than 100 microg must be correctly measured in gravimetric analysis of particles collected on filters. Even small variations in mass measurement may contribute significant errors to calculated concentrations. In addition to the collected particles, a number of other factors affect the observed mass difference between the measurements before and after sampling. The most often controlled of these factors are static charge, temperature, and humidity. Using 951 laboratory blank filter weights, we have statistically analyzed these and other factors that affect the observed filter weight. Some of these are controllable or correctable; others are not and enter into the final results as errors. The standard deviation of differential blank filter weighing after applying all corrections was 2.7 microg. The most important correctable factors are air buoyancy variation and filter storage time. When weighing blank Teflon filters at relative humidity < 50%, these are an order of magnitude more important than weighing-room humidity. Using field blank filters in each weighing batch could control these three factors but also doubles the errors caused by balance random variation and filter handling contamination, because four weighing measurements and the handling of two filters are needed to obtain one corrected differential mass result.