Timber, an ancient material widely used in construction, possesses unique properties but exhibits relative brittleness, particularly in bending. To enhance the performance and reinforce wooden structures, the incorporation of fiber-reinforced polymer (FRP) components has been proposed as a viable solution. However, the timber-FRP system has received less attention compared to the concrete-FRP system in literature, resulting in limited prediction models. Thus, the reliability of these models requires further examination. This research presents an experimental investigation aimed at evaluating the bond strength of timber members strengthened by FRPs, utilizing a pull-out test with various FRP sheets and adhesives. The acquired data is combined with existing experimental data from the literature to propose a novel probabilistic regression model. Furthermore, a comprehensive reliability analysis of the timber-FRP system is conducted by gathering models previously presented by researchers. These models are then compared with the newly developed probabilistic models. Considering three loading levels and defining uncertainty in six influential variables, 192 first-order reliability analyses are performed in two scenarios: one incorporating the model factor and the other without it. The results reveal a significant decrease in the reliability index when uncertainties are considered during member design. The existing models exhibit an average reliability index decrease from 4.26 to 2.82. Additionally, as anticipated, the influence of the live/dead load ratio on determining the reliability index diminishes in the presence of uncertainties.