Austenitic stainless steels is used in many components of nuclear power plants, particularly in the pipes of cooling systems. Owing to power transients and to start-ups and shutdowns, these components are subjected to thermo-mechanical loadings (low-cycle fatigue LCF & high cycle fatigue HCF) and flow-induced vibration (HCF). The corresponding fatigue design curves were established with solid smooth specimens tested in air at room temperature under strain-control. However, these curves do not consider the influence of LWR water environments that were shown to reduce fatigue life. US NRC Regulatory Guide 1.207 or other national equivalents (JSME Code in Japan) then were established taking strain rate, dissolved oxygen and temperature into consideration. Besides, a significant number of issues have been recently identified, which may have an impact on fatigue life but have not been sufficiently investigated, e.g., the potential negative effects of mean stress, mean strain, surface finish, long-term static hold, specimen geometry, multiaxial stress state were sensitivities not explicitly addressed. This study was launched to assess the effect of mean stress on fatigue behavior of a 316L austenitic steel in boiling water reactor environment with hydrogen water chemistry (BWR/HWC) at 288°C. Load-controlled fatigue tests were selected as the easiest experimental technique to impose a pre-defined mean stress. The tests were carried out on hollow specimens at different stress amplitudes and mean stresses in BWR/HWC environment and in air at 288°C, the latter serving as reference environment to evaluate the life reduction induced by the BWR/HWC medium. Few tests in strain-controlled were performed to derive a consistent way to represent the data obtained with the two control modes together. The test results reported in the form of stress-life curves showed that, in LCF regime (< 1e5 cycles), positive and negative mean stresses increase and BWR/HWC environment decreases fatigue life. In the HCF regime, negative mean stress is always beneficial for fatigue life but positive mean stress decreases fatigue life in BWR/HWC. The beneficial effect of mean stress is attributed to its enhanced cyclic hardening on material, which leads to smaller strain amplitude at given stress amplitude. The load-controlled data were converted into strain-life presentation by considering the average strain amplitude over the whole loading cycles. Additionally, a modified Smith-Watson-Topper mean stress correction method was successfully used to correlate the results with and without mean stress. Finally we showed that strain energy density criteria are good alternatives to correlate all data. Crack growth rates were determined from the striation spacing on the fracture surfaces with high resolution scanning electron microscopy (HRSEM); crack initiation sites were also studied with HRSEM, and the microstructures were observed with transmission electron microscopy and electron channeling contrast imaging. From the striation spacing measurement, we concluded that BWR/HWC environment essentially reduces the number cycles needed to initiate a physical crack (crack depth < 50 microns) but modifies slightly the crack growth rate, which correlates well with a strain intensity factor and J-integral method. Life predictions were also done by a well-trained artificial neuron network that considers mechanical, environmental and material properties factors.