Exploring spatiotemporal dynamics and influencing factors of stem respiration (RW) is important for estimating and modeling forest carbon budgets, and helpful to understand variability of forest productivity. In this study, we in situ measured the RW of four major tree species in the temperate forests of northeastern China. The species were Korean pine (Pinus koraiensis), Dahurian larch (Larix gmelinii), white birch (Betula platyphylla), and Manchurian ash (Fraxinus mandshurica). Our objective was to examine diurnal courses of RW across the growing season in order to understand temporal patterns and influencing factors of RW and temperature coefficient of RW (Q10) for the four species. For each species, nine trees were randomly sampled to cover the distribution range of the tree diameters. A polyvinyl chloride collar (inner diameter 10.2 cm, height 5 cm) was cut and polished to fit the stem shape of each sample tree, and installed on the northward side at breast height. The collar was attached with silicon adhesive to the stem surface that was pretreated without causing any injury of the live tissues, and was kept continuously throughout the measuring period. Each month from May to September of 2008, the RW and stem temperature at 1 cm depth under bark (TW ) were measured every two hours with a Li-6400 infrared gas analyzer and digital thermometer. The measured RW was calibrated with the collar volume and stem surface area of each sample tree. Tree species, diameter at breast height (DBH), month and their interactions significantly influenced the RW. The grand means of RW during the measuring period for the larch, ash, birch, and pine were 3.69 μmol?m-2?s-1, 3.24 μmol?m-2?s-1, 1.64 μmol?m-2?s-1, and 1.62 μmol?m-2?s-1, respectively. The diurnal courses of RW for all tree species during the growing season (except for July) overall showed a “unimodal” pattern, largely in accordance with those of TW. However, the occurring time of the peak RW delayed to that of the TW by 2-6 hours depending on species and month. In July, the diurnal course of RW displayed a “sinusoidal” or non-peak pattern for all species. The seasonal variation of daily mean RW showed a bell-shaped curve pattern, with its maximum occurring in July for all species, which was in consistent with the seasonality of temperature and penology. There was a consistent seasonal pattern of Q10 for all tree species, with its minimum occurring in August and maximum in September. The Q10 was significantly different among tree species, varying from 1.09 for the pine and 2.95 for the larch. The mean RW for each individual tree within a specified species was significantly positively correlated with DBH for all species except for the birch. However, the model form and variability of these correlations varied with tree species. It was concluded that the temporal and inter-specific variations of stem respiration and its temperature sensitivity should be considered in developing mechanistic models of forest carbon cycles.