Given a Boolean function f:{0,1}n→{0,1}, the goal in the usual query model is to compute f on an unknown input x∈{0,1}n while minimizing the number of queries to x. One can also consider a "distinguishing" problem denoted by f: given an input x∈f−1(0) and an input y∈f−1(1), either all differing locations are replaced by a ∗, or all differing locations are replaced by †, and an algorithm's goal is to identify which of these is the case while minimizing the number of queries. Ben-David and Kothari [ToC'18] introduced the notion of randomized sabotage complexity of a Boolean function to be the zero-error randomized query complexity of f. A natural follow-up question is to understand (f), the quantum query complexity of f. In this paper, we initiate a systematic study of this. The following are our main results: ∙ If we have additional query access to x and y, then (f)=O(min{(f),n‾√}). ∙ If an algorithm is also required to output a differing index of a 0-input and a 1-input, then (f)=O(min{(f)1.5,n‾√}). ∙ (f)=Ω((f)‾‾‾‾‾‾√), where (f) denotes the fractional block sensitivity of f. By known results, along with the results in the previous bullets, this implies that (f) is polynomially related to (f). ∙ The bound above is easily seen to be tight for standard functions such as And, Or, Majority and Parity. We show that when f is the Indexing function, (f)=Θ((f)), ruling out the possibility that (f)=Θ((f)‾‾‾‾‾‾√) for all f.