To address the effectiveness of predawn leaf water potential in plant‐based irrigation scheduling, Gossypium hirsutum L. plants were grown under fully irrigated and dryland conditions and under three predawn water potential (Ψ_PD) thresholds (−0.5, −0.7, and −0.9 MPa). Measurements included Ψ_PD, plant height, mainstem node number, lint yield, water productivity, and continuous crop canopy temperature. We found that Ψ_PD produced similar yields to current practices, while decreasing overall water use from 7 to 31%, depending on rainfall levels and the treatment utilized. When considered across both years of the study (2013 and 2014), using a −0.5‐MPa Ψ_PD irrigation threshold consistently resulted in less irrigation applied than the checkbook method and maximum water productivity and lint yield. Using a well‐watered baseline developed in 2013 for canopy temperature versus vapor pressure deficit, we calculated a crop water stress index (CWSI) that exhibited a very strong, nonlinear relationship with season average Ψ_PD values between approximately −0.4 and −0.7 MPa (r² = 0.81). A strong, nonlinear relationship was also seen between CWSI and lint yield (r² = 0.81). Predawn water potential appears to be an effective means of determining the need for irrigation in cotton, and in the current study, yield and water productivity were maximized at a season‐long average Ψ_PD threshold of −0.5 MPa. Furthermore, when calibrated using Ψ_PD–based irrigation triggers, canopy‐temperature derived CWSI appears to be a promising tool for future automated plant‐based irrigation scheduling in the southeastern United States.