This in vitro study evaluated the quantitative microleakage (ML) of (A) Gluma/Pekafill, (B) Tripton/Opalux, (C) Syntac/Heliomolar, (D) Prisma Universal Bond 2/Prisma AP.H., (E) Prisma Universal Bond 3/Prisma AP.H., and (F) Clearfil PhotoBond/Photo Clearfil Bright to cementum (dentin) by means of a dye-recovery spectrophotometric technique. Standardized cylindrical Class V preparations, 3 mm in diameter and 1.5 mm deep, were made on the facial surfaces of the roots of ninety extracted human maxillary and mandibular permanent canines and the teeth were restored with the recommended resin composite. The root apices were removed, Class V preparations made at the cut apices and restored with two coats of Copalite varnish and amalgam. The teeth were painted with two coats of nail varnish except for 1 mm surrounding the restorations prior to thermocycling x500 in 2% methylene blue solution. Root sections including the restorations were placed in 50% HNO3, the dye extracted and the dye concentrations determined spectrophotometrically. Standard solutions with known amounts of dye were prepared in 50% HNO3 and unrestored root sections placed in these. ML was expressed as micrograms dye/restoration and data analyzed by ANOVA and Duncan's test. The following results were obtained: (A) 3.67 +/- 2.18; (B) 2.76 +/- 1.71; (C) 2.57 +/- 0.97; (D) 2.56 +/- 2.51; (E) 1.21 +/- 0.24; (F) 0.49 +/- 0.20. Clearfil PhotoBond and Prisma Universal Bond 3 showed significantly less microleakage than other systems evaluated.