In-cylinder flows in internal combustion (IC) engines have always been a focus of study in order to gain better understanding of fuel–air mixing process and combustion optimization. Different conventional experimental techniques such as hot wire anemometry (HWA), laser Doppler anemometry (LDA), and numerical simulations have been grossly inadequate for complete understanding of the complex 3D flows inside the engine cylinder. In this experimental study, tomographic particle imaging velocimetry (PIV) was applied in a four-valve, single-cylinder optical research engine, with an objective of investigating the in-cylinder flow evolution during intake and compression strokes in an engine cycle. In-cylinder flow seeded with ultra-fine graphite particles was illuminated by a high energy, high frequency Nd:YLF laser. The motion of these tracer particles was captured using two cameras from different viewing angles. These two-directional projections of flowfield were used to reconstruct the 3D flowfield of the measurement volume (36 × 25 × 8 mm3), using multiplicative algebraic reconstruction technique (MART) algorithm. Captured images of 50 consecutive engine cycles were ensemble averaged to analyze the in-cylinder flow evolution. Results indicated that the in-cylinder flows are dependent on the piston position and spatial location inside the engine cylinder. The randomness of air-flow fields during the intake stroke was very high, which became more homogeneous during the compression stroke. The flows were found to be highly dependent on Z plane location inside the engine. During the intake stroke, flows were highly turbulent throughout the engine cylinder, and velocities vectors were observed in all directions. However, during the compression stroke, flow velocities were higher near the injector, and they reduced closer to the valves. Absolute velocity during compression stroke was mainly contributed by the out of plane velocity (Vz) component.