The present work aimed at evaluating the use of the LERCAFÉ test to estimate germination and characterize mechanical damage in coffee seeds. The experiment was conducted in the Laboratory of Seed Research at the Universidade Federal de Viçosa, Viçosa-MG. Coffea arabica seeds were submitted to the following treatments: Lot 1 (0% of damaged seeds), Lot 2 (5% of damaged seeds), Lot 3 (10% of damaged seeds) and Lot 4 (15% of damaged seeds). The mechanical damage was caused with the use of the Pulverisette 14 (Fritsch) apparatus. The damage was caused randomly and individually in the seeds, and the lots proportionally combined undamaged and damaged seeds, totaling 1600 seeds per lot. The seeds were evaluated by the germination and LERCAFÉ tests. For this type of damage, the germination results achieved by the LERCAFÉ test presented high correlation with those achieved by the germination test (r=0,9550). Although being randomly caused in the seeds, in other words, without reaching a pre-determined region, the mechanical damage was always characterized by a breach in the embryo region and/or in the region opposite to the embryo and the edges of these breaches presented a green stain. The LERCAFÉ test was efficient to estimate germination and characterize mechanical damage in coffee seeds. Lots of seeds mechanically damaged presented a sharp decrease in the germination power.