Camera traps are often used to monitor wildlife occupancy and, increasingly, to record life history events. For tree‐denning species such as fishers (Pekania pennanti), camera traps have been used to investigate reproduction (e.g., litter size and kit survival) by documenting females with kits as they move between trees. Yet, unbiased quantification of litter size and kit survival likely requires consideration of detection probability during transitions. We conducted a field experiment with the objective of estimating detection probability of fishers coming to or going from a central tree given different camera arrays and compared that to observations of collared females denning in northern New York. We estimated detection probability via logistic regression and investigated the effects of movement differences related to leaving vs. returning to the tree, as well as temporal effects related to kit development. The number of cameras deployed strongly influenced detection probability, which on average ranged from 0.21 with one camera to 0.85 with 6 cameras, yet detection rate of fishers returning to the den was consistently higher than fishers leaving. Despite the difference in detection relative to direction, we obtained at least one photo of 94% of expected kit transfers. Our results demonstrate the benefits and challenges of purpose‐specific tests when using camera traps for estimating demographic rates, and we offer considerations for researchers that may be considering trade‐offs between number of cameras and different collar technology in understanding reproductive ecology of fishers and similar arboreal‐denning carnivores.