BACKGROUND: Increased attention allocation to negative-valenced information and decreased attention allocation to positive-valenced information have been implicated in the etiology and maintenance of depression. The Matrix task, a free-viewing eye-tracking attention assessment task, has shown corroborating results, coupled with adequate reliability. Yet, replication efforts are still needed. Therefore, we replicated a previously published study in depression, using the same task and attention measures. We also explored the potential added effect of depression history on attention allocation. METHODS: Participants diagnosed with major depressive disorder (n = 65) and a matched control group of healthy participants (n = 37) freely viewed 60 different face matrices, each presented for six seconds and comprised of eight sad and eight happy faces. Attention allocation to corresponding areas of interest (AOIs) was compared, and the internal consistency of attention allocation measures was assessed. We then compared the attention allocation of participants amidst their first episode (n = 33) to that of participants with a recurrent depressive episode (n = 32). RESULTS: A significant group-by-stimulus type (happy vs. sad faces) interaction emerged for total dwell time, replicating the findings of the original study. Groups differed on attention allocation to both the sad and happy faces. No findings emerged for first fixation measures. Internal consistency of the total dwell time measure was high. Depression history had no effect on attention allocation. LIMITATIONS: Due to ethical constraints (delay of treatment), test-retest reliability was not assessed. CONCLUSIONS: The Matrix task provides a reliable and replicable measure of attention allocation in MDD, showing no effects for depression history.