ABSTRACT: Fast acquisition GPS technologies such as Fastloc GPS have been commonly used in recent years to study fine scale spatio-temporal ecology of marine vertebrates. While Fastloc GPS gives more accurate locations than earlier methods, it remains important to identify and remove locations with high error because some location fixes are much less accurate than others. Increasing the number of source satellites required for a valid fix is a simple filter method but it comes at the cost of great data loss. Using data sets acquired from loggerhead turtles Caretta caretta, we explored an alternative filtering approach, based on speed between successive locations, angles created by 3 consecutive locations, manufacturer’s quality index, and number of satellites used for location calculation. The performance of the proposed filter method was evaluated by conducting terrestrial, mobile tests. When our filter method was used, the linear error (mean ± SD) of Fastloc GPS data decreased from 2645.5 ± 29458.2 m (n = 1328) to 47.1 ± 61.0 m (n = 1246), while retaining more than 94% of data. Our filter method also led to more accurate home range estimates than the simple filter method. Improvements in data retention and home range estimates will give more reliable information for marine spatial planning and habitat protection.