No. A SMART does not have to include an embedded tailoring variable. It depends on the study’s scientific questions, including what is already known about the adaptive intervention being optimized. An embedded tailoring variable is often used in settings in which it is already known that subsequent stage intervention options should not be provided (or only be provided) to a subgroup of individuals based on the values of the tailoring variable. For example, the SMART below has no embedded tailoring variable. All participants in this SMART are randomized in stage 1 to MSM or BPT; and all participants are again randomized in stage 2 to continued treatment or MSM+BPT.  In fact, one of the goals of this SMART is to learn how best to define tailoring variables. Recall that SMARTs are used to provide evidence for any component of an adaptive intervention.