Visual features previously associated with reward can automatically capture attention even when task-irrelevant, no longer rewarded and not physical salient. This phenomenon is known as value-driven attention capture (Anderson et al., 2011b). The large majority of VDAC research uses a largely homogenous task and it remains unclear whether VDAC can generalize outside of this paradigm. This work aimed to investigate the generalizability of VDAC across three different factors: time, domain, and environment. VDAC has been shown to persist for a prolonged time after learning reward associations (Anderson, 2013; Anderson et al., 2011b). However, the apparent persistence is observed within tasks that have consistent features and it is unclear which of these features are necessary to observe the persistence of this phenomenon over time. In six experiments, these task features were manipulated including the task-relevance of the reward-associated feature, absent trials where no reward associated feature is present, and a target defined by a salient shape. It was found that VDAC was only reliably observed when replicating typical design features where there is the inclusion of absent trials and the target is defined as a salient shape singleton, factors that are known to influence search (Geyer et al., 2008; Lamy & Egeth, 2003). The second aspect of generalizability that was assessed was attentional domain, VDAC is often observed in a task that requires spatial selective attention. In three Experiments, the effects of VDAC were observed during a different type of attention, sustained attention. Results showed that when the reward-associated feature remained task-relevant there was a profound effect of reward on performance, such that participants showed a benefit on rewarded relative to non-rewarded trials. However, there was no impact of the reward-associated feature when it was rendered task-irrelevant at test suggesting that VDAC may not generalize to sustained attention when not relevant to the current goals. Finally, in two experiments the generalizability of VDAC across different environments was explored. In the first experiment, it was found that reward-associations learned in the typical VDAC search paradigm influenced performance in an unrelated task one-week later indicating that VDAC can generalize even when the task requirements and attentional set differ between learning and test. In Experiment 2 the generalizability of VDAC was assessed using eye-movements while interacting with real-world objects. An effect of reward history was not found indicating varied findings with regard to the generalizability of VDAC. Although, an effect of selection history was observed suggesting that selection history may have a stronger role in the real world when initiating movements to act upon objects.