Residual stresses are an integral part of the total stress acting on any component in service. It is important to determine and/or predict the magnitude, nature and direction of the residual stress to estimate the life of important engineering parts, particularly welded components. This work aims to introduce experimental models to predict residual stresses in the heat-affected zone (HAZ). These models specify the effect of laser welding input parameters on maximum residual stress and its direction. The process input variables considered in this study are laser power (1.03 - 1.368 kW), travel speed (26.48 – 68.52 cm/min) and focal point position (- 1 to 0 mm). Laser butt-welding of 304 stainless steel plates of 3 mm thick were investigated using a 1.5 kW CW CO2 Rofin laser as a welding source. Hole-drilling method was employed to measure the magnitude, and direction of the maximum principal stress in and around the HAZ, using a CEA-06- 062UM-120 strain gauge rosette, which allows measurement of the residual stresses close to the weld bead. The experiment was designed based on Response Surface Methodology (RSM). Fifteen different welding conditions plus 5 repeat tests were carried out based on the design matrix. Maximum principal residual stresses and their directions were calculated for the twenty samples. The stepwise regression method was selected using Design-expert software to fit the experimental responses to a second order polynomial. Sequential F test and other adequacy measures were then used to check the models adequacy. The experimental results indicate that the proposed mathematical models could adequately describe the residual stress within the limits of the factors being studied. Using the models developed, the main and interaction effect of the process input variables on the two responses were determined quantitatively and presented graphically. It is observed that the travel speed and laser power are the main factors affecting the behavior of the residual stress. It is recommended to use the models to find the optimal combination of welding conditions that lead to minimum distortion.