Introduction: On the one hand, Cognitive Load Theory effects are obtained when learning and processing materials are demanding. When they are less demanding, when the learners have more knowledge, then an expertise reversal effect is obtained. What works for novices does not work for experts nor for advanced learners. In other words, improving learning by decreasing cognitive demand is obtained if and only if cognitive demand is high. On the other hand, there is a lot of literature about desirable difficulties, how it works and when it happens. Introducing difficulties in the material or introducing an additional task can improve learning. Literature in this domain shows that desirable difficulties effect is obtained when these difficulties are relevant, mainly when difficulties increase cognitive engagement in the learning task, when they result in deeper processing of the learning material (Bjork, 2013). On the contrary, when the difficulties are not relevant (e.g. Diemand-Yauman, Oppenheimer, & Vaughan, 2011) the desirable difficulties effect is hard to obtain, and when obtained, it is hard to replicate (Yue, Castel & Bjork, 2012). Chen, Castro-Alonso, Paas and Sweller (2018) recently argued that “many failures to obtain desirable difficulty effects may occur under conditions where working memory is already stressed due to the use of high element interactivity information. Under such conditions, the introduction of additional difficulties may be undesirable rather than desirable”. These authors obtained empirical results that confirm this hypothesis. Our hypothesis is basically the same, but more general: Desirable difficulties effect is obtained when learning and materials are very easy to process, for any kind of reason; exactly like with the expertise reversal effect that is obtained when the material is easy to process for any kind of reason, i.e. for any cognitive load theory effect. Method: We are reviewing empirical papers (25 papers from cognitive load theory, 25 papers from desirable difficulties literature). We selected papers where the performance at the post-test is given as a rate or percentage of success. When the control group obtained a high performance, we considered that the material was not demanding. When the control group obtained a low performance, we considered that the material was demanding. Then we calculated the rate of experiments where a desirable difficulty effect was obtained and the material was not demanding, divided by the total number of experiments where a desirable difficulties effect was obtained. In the same way, we calculated the rate of experiments where a CLT effect was obtained and the material was demanding, divided by the total number of experiments where a desirable difficulties effect was obtained. Results: The papers we already read confirm that desirable difficulty effect is obtained if and only if post-tests in the control group are performed at a very high level. Papers collection and data analysis are ongoing. Discussion: The fact that many authors forget to check if the material was demanding or not could explain why it is difficult to replicate the desirable difficulties effect.