Radiocollars are an increasingly important tool in wildlife research. Yet, as with all remotely compiled data, measurement error is inherent in the technology. We directly compare radiocollars with low measurement error (Global Positioning System [GPS]) with radiocollars with high measurement error (Argos satellite). Specifically, we compare how differences in precision between GPS and Argos satellite technologies affect the estimation of resource selection functions (RSFs). We estimated RSF models from GPS and Argos satellite radiocollar data collected in December 2008 through April 2009 from wolves within the same pack in southwestern Alberta, Canada, and used Akaike's Information Criterion (AIC) to identify the most parsimonious models. In general, β coefficients were closer to zero and coefficients of variation were higher for models estimated using Argos data. But even more serious, AIC identified different top models between the Argos and GPS data sets because measurement error alone can induce attenuation bias, which leads to erroneous conclusions on selection of habitats. GPS radiocollar data were more precise and more accurate, resulting in RSF models that were a better representation of true habitat selection by each wolf pack. © 2013 The Wildlife Society.