Tides are often the largest source of sea levels fluctuations. Two new probabilistic tsunami hazard assessments (PTHA) methods are proposed to combine the tidal phase uncertainty at the moment of tsunami occurrence with other sources of uncertainty. The first method adopts a Stochastic Reduced Order Model (SROM) producing sets of tidal phase samples to be used in tsunami simulations. The second method uses tsunami simulations with prescribed collocation tidal phases and tide probability distributions to model the uncertainty. The methods are extended to non-stationary probabilistic tsunami hazard assessment, compounding tsunamis, tides and sea level rise (SLR). As an illustration, these methods are applied for assessing tsunamis generated in the Manila Subduction Zone, on the coasts of Kao Hsiung and Hong Kong. While the SROM-based method is faster solving for the PTHA if only tides are considered, the collocation-based method is faster when both SLR and tides are considered. For the illustration case, tides have a relevant impact on PTHA results, however, the SLR within an exposure time of 100 years has stronger impact. PTHA curves of the maximum tsunami elevation are affected by tides and SLR differently. While tides and SLR increase the dispersion of PTHA hazard curve distributions, the latter also produces a translation toward higher elevations. The development of formulations based on SROM or collocation tides is the key to establishing a method which feasibly can be applied to other regions for comprehensive analysis at a global scale.