Current knowledge on crop irrigation in greenhouses is limited, and most irrigation measures are based on evapotranspiration (ET) estimated using external climatic conditions. Moreover, crop water requirement estimation by the Penman-Monteith (PM) model assumes complete coverage of the soil by the plants and usually does not account for partial cover, which changes during the growth of annual crops. The main goals of the present research were: (i) to develop and apply irrigation demand estimates for greenhouse pepper crops during their growth, based on internal meteorological conditions and real-time plant characteristics; and (ii) to examine the effect of that irrigation on yield. Experiments were conducted during two consecutive seasons in greenhouse pepper crops in the Jordan Valley region of eastern Israel. Four irrigation treatments were applied based on four different ET models: (i) PM model for outside reference evapotranspiration based on the FAO56 methodology, also denoted as the commercial treatment; (ii) PMrb model for pepper plants, which is a standard PM model in which a boundary layer resistance replaces the aerodynamic resistance; (iii) Shuttleworth-Wallace Layer (SWL) model which considers the soil and plants as separate interactive layers; (iv) Shuttleworth-Wallace Patch (SWP) model which considers the soil and plants as independent patches. The commercial treatment used external climatic conditions from a nearby meteorological station, whereas the other three treatments used internal meteorological conditions measured inside the greenhouse. Results showed a high correlation (R2 > 0.88) between leaf area index, vegetation cover fraction, and plant height during plant growth. Using the same meteorological conditions inside the greenhouse, the seasonal ET estimated by the PMrb model, which considers full vegetation cover, was larger by 21-28% than the SWL and SWP models, which consider the partial plant cover. Irrigation doses were deduced from ET multiplied by crop coefficients derived in a previous study for the same crop and structures. Using average data of the two seasons, we found that seasonal irrigation based on the two SW models was lower by 5-15% than that based on PMrb, and by 35-42% than the commercial irrigation. The water use efficiency of the commercial treatment was the lowest, while for the other treatments it was about 0.02 ton m-3 and higher by about 40% than the commercial. We conclude that irrigation of annual crops based on partitioning ET into evaporation and transpiration based on crop coverage and using internal climatic conditions allows water-savings without significant yield loss.