Abstract: Venturi-type bubble generator as well as its performance has been received wide attentions by engineers and researchers in recent decades. In this paper, experimental research was carried out to study flow regime and pressure drop of the gas-liquid (air-water) twophase through a Venturi bubble generator. The bubble generator was designed with a rectangular cross section, and the gas was injected into the high speed water flow from the throat section of the Venturi. Interface distribution and dynamic processes of bubbles were observed with a high speed camera, and the static pressures of inlet, throat and outlet of the bubble generator were measured by pressure sensors. Such three basic regimes as bubbly, slug and column are preliminarily identified in the bubble generator by visual experimental observations. A significant increase of the pressure drop of the bubble generator is found in the experiment when the flow regime transformed from slug into column. In order to identify the flow regime of the bubble generator, time frequency of pressure signal was analyzed, and it is found that the outlet section of the bubble generator is confirmed as the most favorable position for pressure signal on-line monitoring. A relative kurtosis of probability density function and a coefficient of variation of power spectrum density strongly associated with flow regime transition were extracted from the pressure signals, and they were applied to identify slug-column and bubbly-slug transition, respectively. Since the traditional pressure drop models can’t precisely predict the whole pressure drop of the Venturi bubble generator, a new pressure drop prediction method is proposed. The effects of partial liquid single-phase flow and flow regime transition on the pressure drop prediction are considered in the new correlation by introduced two modified parameters, ′ and θ′, respectively. Both of the two modified parameters were fitted from the current experimental results. Prediction results of the new correlation agree well with those of the experiments with a relative root mean square deviation of 10.74%. The new correlation is applied to the two-phase flow under conditions of normal temperature and 1-1 times of atmospheric pressure, while the liquid flow Reynolds number ranges from 1.7×104 to 7.0×104 and the low mass quality ranges from 0 to 5.1×10-5. The new method proposed in this paper for predicting pressure drop of the device regarding gas-liquid two-phase flow can provide theoretical bases and references for the online monitoring of the similar device.