Bidirectional photobiomodulation (PBM) therapy is an active research area. However, most studies have focused on its dependence on optical parameters rather than on its tissue-dependent effects. We constructed mouse models of wounds in three inflammatory states (normal, low, and high levels of inflammations) to assess the bidirectional regulatory effect of PBM on inflammation. Mice were divided into three groups to prepare common wounds, diabetic wounds, and bacteria-infected wounds. The same PBM protocol was used to regularly irradiate the wounds over a 14 d period. PBM promoted healing of all three kinds of wounds, but the inflammatory manifestations in each were significantly different. In common wounds, PBM slightly increased the aggregation of inflammatory cells and expression of IL-6 but had no effect on the inflammatory score. For wounds in a high level of inflammation caused by infection, PBM significantly increased TNF-α expression in the first 3 d of treatment but quickly eliminated inflammation after the acute phase. For the diabetic wounds in a low level of inflammation, PBM intervention significantly increased inflammation scores and prevented neutrophils from falling below baseline levels at the end of the 14 d observation period. Under fixed optical conditions, PBM has a bidirectional (pro- or anti-inflammatory) effect on inflammation, depending on the immune state of the target organism and the presence of inflammatory stimulants. Our results provide a basis for the formulation of clinical guidelines for PBM application.