The purpose of this study was to evaluate the influence of surface treatments and thermocycling on the microtensile bond strength (μTBS) of self-adhesive resin cement to human enamel and dentin. Eighty human third molars were selected. The crowns of 40 teeth were transversally sectioned, exposing the mid-coronal dentin. The buccal surfaces of the other 40 teeth were grinded to obtain a 5 mm2 flat enamel area. Eighty resin blocks were produced and cemented to the dental surfaces with RelyX Unicem, then grouped according to the surface treatment (n = 10): UnicemC with no conditioning, UnicemP with 37% phosphoric acid/15 s, and UnicemPA with 37% phosphoric acid/15 s plus adhesive bonding (Single Bond 2). There were two control groups, one for enamel and the other for dentin: VR with 37% phosphoric acid/15 s plus adhesive bonding (Single Bond 2) plus Variolink II. The enamel-dentin resin cement blocks were sectioned to produce non-trimmed bar specimens, which were divided into two storage conditions: dry, μTBS immediately after cutting; TC (5,000 x; 5°C/55°C). The samples were submitted to μTBS, and data were statistically analyzed by ANOVA and Tukey's test. The results showed statistical differences between UnicemC and the others. UnicemPA and VR showed better bond strength to dentin during the period before and after thermocycling, respectively. For the enamel, UnicemP showed better bond strength for both situations. Only for UnicemPA did the thermocycling significantly decrease the bond strength values. Within the limits of this study, it could be concluded that the bond strength is influenced by the surface treatments, and that thermocycling decreases the bond strength of all groups, but significantly only for UnicemPA.