This study presents fatigue test results of longitudinal stiffener specimens in as-welded and high frequency mechanical impact (HFMI) post-treated condition under both constant amplitude loading (CAL) and variable amplitude loading (VAL). The experiments incorporate mild steel S355 and high-strength steel S700 samples at load stress ratios of R = 0.1 and R = −1 respectively. VAL tests are performed utilizing a straight-line distribution with a sequence length of 2 ⋅ 105 load-cycles and an omission level of 19% of the maximum stress range. The test results reveal a significant benefit of the HFMI-treatment under CAL. On the contrary, the increase in fatigue strength is majorly reduced in case of VAL. X-ray residual stress measurements for the investigated mild steel S355 specimens demonstrate that even at comparably minor load stress ranges, a certain relaxation of the HFMI-treatment induced local compressive residual stress state at the weld toe occurs. This phenomenon contributes to the decrease of the benefit by the HFMI-treatment under VAL compared to CAL. An evaluation of specified damage sums on the basis of the determination of an equivalent stress range for VAL reveals that damage sums between D = 1.0 and D = 0.5 are well applicable for a conservative fatigue assessment on the basis of the IIW-recommended fatigue design curves. However, facilitating a lightweight component design by considering experimental test results of HFMI-treated joints under CAL, a reduced specified damage sum of D = 0.3, or even of D = 0.2 in case of fluctuating mean stress states, is suggested for the fatigue assessment to avoid non-conservative results.