Whereas helping is costly for the helper, it is beneficial for the person who requests help. However, there is only scarce evidence on the relative costs and benefits of helping and this evidence is mixed. In addition, hardly any research investigates how these costs and benefits can be manipulated. With a laboratory experiment, we first examined how helping affects the performance of the helper, the help requester, and the dyad. Second, we investigated whether quiet hours that structure time into spans with interruptions and spans without interruptions decrease the costs of helping while keeping its benefits. We found that the requester's performance was higher and the helper's performance lower when help requests were permitted at any time rather than when no help was allowed. However, overall performance fell short of being significantly higher with help at any time. In addition, the helper's performance failed to be higher with quiet hours compared to interruptions at any time. Instead, both the helper's and the requester's performance were lower with quiet hours, resulting also in a lower overall performance. In search of an explanation, our data indicate that structuring time into spans with and without interruptions might generate costs of their own that could be reduced by setting fewer but longer spans without interruptions.