Several studies have explored the effects of divided attention on priming, but little is known about the impact of working memory load on implicit visual memory. The aim of this study was to determine whether there are differential effects of working memory load on a visual priming task compared to a recognition task. In the encoding phase, participants were presented with real-object pictures and asked to classify them semantically. At retrieval, 40 studied and 40 new images were presented (partially masked) for 100 ms, and participants had to identify the object. Each trial was immediately followed by a recognition test, in which the unmasked image was shown again, and participants had to indicate whether it had been presented at encoding or not. Regarding working memory load, participants performed a task in which a load was imposed in half of the trials. Twenty-four participants concurrently performed an articulatory suppression task, another group of 24 subjects performed an executive demanding task, and a third group of 24 participants performed a spatial tapping task. Working memory load failed to diminish performance on both priming and recognition tests in the articulatory suppression condition. However, the backward counting and the tapping tasks influenced recognition, rather than priming. The relative pattern of backward counting effects on recognition and priming were then broadly replicated in a follow-up experiment using an adapted priming task (N = 24). Results suggest that a concurrent load has a more robust effect on recognition than on priming, especially when the working memory task is executively demanding.