The Global Positioning System (GPS) uses static relative positioning to mitigate common errors and has the ability to remove the effects of ocean tidal loading (OTL). OTL effects cannot be completely removed by increasing the baseline length during GPS data processing and must be corrected with OTL models, especially in Taiwan, which is surrounded by the ocean. In this paper, we collected GPS data from ten stations that are located along the coast of Taiwan. The periodic effect of OTL is discussed through a wavelet analysis and a harmonic analysis. We also calculated the amplitudes and phases of the semi-diurnal and diurnal tides to understand the effect of OTL on the GPS heights along the coast of Taiwan. The GPS data were divided into short-term (1 month) and long-term (2 years) categories and were calculated using the Bernese 5.0 software. The tidal data and the GPS heights were transferred into a wavelet spectrum for analysis. We found that in the tidal data, the periodic M2 signal was stronger than the K1 signal, which was not the case for the GPS height data. This result indicated that the common errors were not completely removed from the GPS heights and that the corresponding period was approximately 1 day, which might result in a stronger K1 signal. Moreover, five OTL models were chosen and compared in this case. Although the GPS heights with different OTL corrections were very similar, NAO.99b was the best fitting model for the ocean around Taiwan. Finally, we analyzed the tidal data and the M2 amplitudes and phases of the GPS heights using a harmonic analysis and compared the results with the NAO.99b model. The comparison indicated that the best agreement occurred in southwestern Taiwan and that the poorest agreement occurred in southeastern Taiwan with a maximum error of 0.3 cm. The phase results showed that the OTL model overlapped with the tidal data in the open ocean around central Taiwan, and both the GPS heights and the heights from the OTL model were 1.5 to 2 h earlier than the tidal data.