summary The aims of this study were: (i) to evaluate and compare retention of two-teeth (implant) supported mandibular overdenture with either stud or magnetic attachments during linear (axial) and rotational (paraxial) dislodgements; (ii) to compare retentive properties before and after wear simulation. The test group consisted of five magnetic and four stud overdenture attachments (n = 12 specimens for each attachment type). Retention in axial direction was evaluated on one-tooth (implant) model by measuring maximum retentive force (N) and range of retention (mm) during the linear dislodgement. Retention in the paraxial direction was evaluated on mandibular-overdenture model by measuring the maximum retentive force (N) during three types of rotational dislodgements – anterior, lateral and posterior. The minimum number of cycles required to simulate wear was determined by special wear test. Afterwards, the wear was simulated in the test group, and retention in axial and paraxial directions was measured again. Statistical analysis: one-way anova, Scheffe post hoc and paired-samples t-tests (P < 0·05). Initially, studs had higher retention (4–11 N) than magnets (4·5–6 N) in axial direction. After the wear simulation, it had decreased from 76% to 48% for some of the studs and had become similar to the retention of magnetic attachments. Magnets had lower retention range (0·2–0·3 mm) than studs (0·5–1·1 mm). Studs provided similar or higher retention in paraxial directions than magnetic attachments both before and after wear simulation. Retentive properties of magnets decreased mostly with posterior rotational dislodgement. Retentive properties of stud overdenture attachments were less constant.