In advance of precise positioning with phase data, cycle slip detection (CSD) is a basic work that should be implemented in phase data possessing. When the cycle slip occurred, cycle slip repair (CSR) can be taken to rebuild the continuity of phase data. Unfortunately, the large pseudorange errors can contaminate the combinations with the pseudoranges and phases such as the Hatch–Melbourne–Wubbena combination (HMW) and cause false CSD or wrong CSR results. On the other hand, the severe ionospheric time variation can deteriorate the epoch-difference geometry-free phase (GF), and tremendously interfere with the performances of CSD and CSR. To handle the aforementioned limitations, a global position system (GPS) triple-frequency CSR method (GTCSR) is proposed with two efficient treatments: (1) the significant ionospheric variations are corrected, and the influences from the residual ionospheric effects are minimized along with the observational noises; and (2) the impacts of large pseudorange errors are refrained by designing a discrimination function with a geometry-free and ionosphere-free phase to identify the correct cycle slip values. Consequently, CSR tests were conducted with three monitoring stations at different regions. First, during a strong geomagnetic storm, without correcting the ionospheric variation of CSR (WICSR) displayed obvious failures, and many epochs of cycle slip values from WICSR deviated from the known values. However, the results of the GTCSR were correct, and GTCSR presented a higher success rate than that of WICSR. Furthermore, for the real triple-frequency data, by adding gross errors of 2.5 m on all epoch-difference pseudoranges epoch by epoch, the conventional triple-frequency CSR with the optimized combinations (CTCSR) and the CSD with HMW (HMWCSD) showed many mistakes, where the results of CTCSR and HMWCSD on numerous epochs were inconsistent with the actual situations, but the success rate of GTCSR was significantly higher than those of CTCSR and HMWCSD. In summary, in the condition of the cutoff elevation being larger than 10 degrees, improved performances and higher success rates were achieved from GTCSR under environments of large pseudorange errors and severe ionospheric variations.