This paper briefly summarizes the results of tests conducted on 18 full-size push-out specimens and 4 pull-out specimens. The objective of the project was to study the behaviour of headed studs in composite beams with ribbed metal decks perpendicular and parallel to the steel beam. The longitudinal spacing of the headed studs and the rib geometry of the metal decks were the principal experimental parameters.Five of the push-out specimens had solid concrete slabs, in five specimens the ribbed metal deck was placed parallel to the steel beam, and in the remaining eight specimens the metal deck was perpendicular to the steel beam. In most cases 38 mm deck and 16 × 76 studs were used.Test results obtained by the authors appear to indicate that with some exceptions the current Canadian Standards Association and Load and Resistance Factor Design codes are able to predict stud strength correctly for specimens with solid slabs and with parallel ribbed slabs only when failure occurs owing to stud shearing, i.e., when the studs are spaced sufficiently apart. It is recommended that these codes include a provision to check the possibility of concrete-related failures when the longitudinal stud spacing approaches or falls below six times the stud diameter for solid and parallel ribbed slabs.Stud pull-out was the principal mode of failure in the specimens with perpendicular ribbed metal decks. An equation recently proposed by Hawkins and Mitchell appears to underestimate the capacity for specimens with 38 mm deck. For specimens with 76 mm deck, their equation overestimates the stud capacity. The authors have proposed two separate but similar empirical equations for specimens with 38 and 76 mm deck. Key words: composite beam, headed stud, stud shear capacity, stud pull-out failure, minimum stud spacing, perpendicular deck, parallel deck.