Accurate and precise reporting of forest survey locations is required to integrate forest survey data with Geographical Information Systems. The accuracies of five Global Positioning System (GPS) survey methods using a hand-held receiver were tested in a mixed forest of trembling aspen and spruce. Accuracy improved by eliminating positions obtained under poor satellite configurations and by using position averaging methods. Single fix positions, taking as little as two minutes to obtain, yielded better than 100m accuracy more than 80% of the time. Allowing the receiver to continuously collect fixes for 15 to 30 minutes and then averaging the fixes yielded a median position error of 17 m. Sixty one stands representing a diversity of cover types, canopy heights, and crown closure in the Petawawa Research Forest were tested during June and July of 1992 for canopy interference with GPS signals. A GPS position was obtained under the canopy in 74% of the stands. Launches of additional GPS satellites since the summer of 1992 have further improved the probability of obtaining accurate geographical positions under forest canopies. Key words: global positioning system, position accuracy, signal reception, canopy interference