From an intention of summarizing present understandings of positive corona discharges in water and atmospheric pressure gases, we tried to observe streamers in those media by reproducing and complementing previously reported results under a common experimental setup. We used a point-to-plane electrode configuration with different combinations of electrode gap (7 and 19 mm length) and pulsed power sources (0.25 and 2.5 e s duration). The general features of streamers were similar and the streamer-to-spark transition was also observed in both the media. However, in the details large differences were observed due to inherent nature of the media. The measured propagation speed of streamers in water of 0.035 × 10 6 ms −1 was much smaller than the speed in gases (air, N 2 and Ar) from 0.4 to 1.1 × 10 6 ms −1 depending on species. In He the discharge looked glow-like and no streamer was observed. The other characteristics of streamers in gases, such as inception voltage, number of branches and thickness did also depend on the species. The thickness and the length of streamers in water were smaller than those in gases. From the volumetric expansion of a streamer in water after the discharge, the molecular density within the streamer medium was estimated to be rarefied from the density of water by about an order of magnitude in the active discharge phase. We derived also the electron density from the analysis of Stark broadened spectral lines of H and O atoms on the order of 10 25 m −3 at the earlier time of the streamer propagation. The analyzed background blackbody radiation, rotational temperature of OH band emission and population density of Cu atomic lines yielded a consistent temperature of the streamer medium between 7000 and 10 000 K. Using the present data with a combination of the analysis of static electric field and previously reported results, we discuss the reason for the relatively low streamer inception voltage in water as compared to the large difference in the medium density between water and gases.