Current cutoff in lightning channels, which takes place in the development of both intracloud and cloud-to-ground flashes, is still poorly understood. A new evaluation of the conditions leading to current cutoff, and also of the two existing hypotheses of the cutoff mechanism, is the main objective of the paper. We reviewed the literature with results of measurements and modeling of free-burning arcs in a laboratory (the closest analogs of lightning leaders) focusing on the relationship between the internal electric field and current. This relationship governs the leader's behavior in the current cutoff. In our analysis of the mechanisms leading to current cutoff, we identify the two types of current cutoff in lightning channels: the current cutoff in a single, unbranched leader channel, which occurs as the result of reaching the threshold conditions for leader propagation; and the current cutoff in branched leaders, when screening by the leader branches alters the ambient electrical environment, thus diminishing the leader current and causing cutoff at a branching point or at the base of the straight channel that preceded branching. We advance the electrostatic model of the screening effect of branching on current cutoff, introduced by Mazur and Ruhnke (1993), and we provide the evidence of this mechanism from lightning observations. We also critically evaluate the concept of lightning-channel instability, proposed by Heckman (1992), as a suggested mechanism leading to current cutoff. We show that the fundamentals of this concept and therefore the concept in its entirety are invalid.