Contamination particles of controlled size and shape were deposited onto 1.14 cm thick fused silica windows by sputtering Al through a mask. The particles were 1 µm thick circular dots, 10 to 250 µm in diameter. Al shavings were also deposited on the windows to investigate the effects of particle-substrate adhesion. The silica windows were then illuminated repetitively using a 3-ns, 355 nm and an 8.6-ns, 1064 nm laser. The tests were conducted at near normal incidence with particles on the input and output surfaces of the window.During the first shot, a plasma ignited at the metal particle and damage initiated on the fused silica surface. The morphological features of the damage initiated at the metal dots were very reproducible but different for input and output surface contamination. For input surface contamination, minor damage occurred where the particle was located; such damage ceased to grow with the removal of contaminant material. More serious damage (pits and cracks) was initiated on the output surface (especially at 355 nm) and grew to catastrophic proportions after few shots. Output surface contaminants were usually ejected on the initial shot, leaving a wave pattern on the surface. No further damage occurred with subsequent shots unless a shot (usually the first shot) cracked the surface; such behavior was mostly observed at 355 nm and occasionally for large shavings at 1064 nm.The size of the damaged area scaled with the size of the particle (except when catastrophic damage occurred). The onset of catastrophic damage on the output surface occurred only when particles exceeded a critical size. The damage behavior of the sputtered dots was found to be qualitatively similar to that of the shavings. The artificial contamination technique accelerated the study by allowing better control of the test conditions.