Fluid flows characterized by density variations have been studied using the schlieren-particle image velocimetry (PIV) system. The knife-edge location plays a crucial role in determining the system’s sensitivity, which significantly affects the accuracy of the measured quantities. Further, the optimum knife-edge position and the correct combination of image recording speed and interrogation window size are desirable for achieving the most accurate and reliable results. The present paper discusses the above issues on the measured quantities, such as temperature field, local Nusselt number distribution along the conducting walls, average Nusselt number, and velocity field. The experiment is performed to investigate laminar and steady natural convective flow in a water-enclosed cubic cavity with a left hot wall and a right cold wall. The analysis is undertaken for various knife-edge positions (0–90%), different image time separation varying (20–200 ms,) and interrogation window size using two passes varying from W1 = 32 pixels, W2 = 16 pixels to W1 = 128 pixels, W2 = 64 pixels. The results are presented for two distinct Rayleigh number, 1 × 108 and 3 × 108. Three-dimensional simulations have been carried out to check the fidelity of the experiment for Ra = 1 × 108. A high dynamic range of temperature is obtained for the range of knife-edge position in 50–65% while a high-velocity range is realized for knife-edge cutoff of 65% and combination of image time separation of Δt = 100 ms and interrogation window size with two passes of W1 = 64 pixels followed by W2 = 32 pixels.