In teleoperation, haptic feedback allows the human operator to touch the remote environment. Yet, it is only partially understood to what extent the quality of haptic feedback contributes to human-in-the-loop task performance. This paper presents a human factors experiment in which teleoperated task performance and control effort are assessed for a typical (dis-)assembly task in a hard-to-hard environment, well known to the operator. Subjects are provided with four levels of haptic feedback quality: no haptic feedback, low-frequency haptic feedback, combined low- and high-frequency haptic feedback, and the best possible-a natural spectrum of haptic feedback in a direct-controlled equivalent of the task. Four generalized fundamental subtasks are identified, namely: 1) free-space movement, 2) contact transition, 3) constrained translational, and 4) constrained rotational tasks. The results show that overall task performance and control effort are primarily improved by providing low-frequency haptic feedback (specifically by improvements in constrained translational and constrained rotational tasks), while further haptic feedback quality improvements yield only marginal performance increases and control effort decreases, even if a full natural spectrum of haptic feedback is provided.