The constant amplitude fatigue behavior of adhesively bonded pultruded glass fiber reinforced polymer double-lap joints were modeled by a number of conceptually different phenomenological SN (cyclic stress vs. number of cycle) formulations. An extended database containing constant amplitude fatigue data under tension (R=0.1), compression (R=10), and reversed loading (R=1) were analyzed in order to investigate whether or not there exists an appropriate fatigue formulation for accurate modeling of the behavior of the examined joints throughout their lifetime, from the very low-cycle fatigue to the high-cycle fatigue regions. Based on an extensive review, appropriate fatigue formulations that take into account the probabilistic nature of lifetime measurements were selected and their fundamental assumptions were examined. The validity of the statistical assumptions of these models was found to be influenced by the applied loading conditions. The modeling results were similar for all selected fatigue formulations with the derived SN curves exhibiting differences mainly in the low- and high-cycle fatigue regimes. The formulations insensitive to the scatter in the experimental data were found to be the most appropriate models.