Installing conservation fences to prohibit feral animal access to wetlands can become a barrier for non-target species of interest. We collected 161 turtles (Chelodina rugosa, Emydura subglobosa worrelli, Myuchelys latisternum) from twenty floodplain and riverine wetlands during post-wet (June–August) and late-dry season (November– December) surveys (2015–2018) in northern Australia. Wetlands were fenced (150 × 150 mm square, 1.05 m high wire mesh) or unfenced around the wet perimeter. Ninety-seven percent of individuals caught in either fenced or unfenced wetlands had a shell carapace width greater than mesh width, of these 44 (46%) were captured inside fenced wetlands, while 50 were caught in unfenced wetlands. The remaining 35 turtles were smaller than 150 mm and would likely pass easily through fence mesh. Sixty-five turtles partook in a fencing manipulative experiment. Turtles with carapace widths wider than mesh often successfully escaped through fences by lifting one side of their shell and passing diagonally through the mesh. In a second experiment where a piece of vertical wire (1500 × 300 mm) was removed, turtles located ‘gates’ after prospecting and fitting through meshing areas that were too small to pass. Ninety-two percent of turtles were able to locate and pass through gates, while 8% failed to locate a gate after 2 h. Gates applied every 4 m showed an 83% passage rate, every 2 m was 91%, and every 1 m was 100%. Combing field and manipulative experiments revealed that large turtles will prospect and move along a fence until they find suitable passage, which has important consequences when considering that gates could be easily retrofitted to existing sites, as well in new fencing programs, which has enormous positive conservation benefits for turtles in an already challenging and changing floodplain environment.