Groundwater take applicants must provide an assessment of environmental effects which includes an estimate of likely stream depletion caused by the proposed groundwater abstraction. The consideration of stream depletion and subsequent surface water allocation varies across regional councils, however most applications use standard spreadsheet models to calculate stream depletion with the Theis/Jenkins equation or the Hunt (1999), Hunt (2003) equations. These methods are quick and easily repeatable and so have value in providing consistent assessments. However, a desktop review of stream depletion calculations for groundwater takes in the Mataura River above Gore in Southland found that the applications had been inconsistently reviewed in the past; in many cases stream depletion had been underestimated and values entered into the models were poorly justified. For this review, documentation surrounding 85 groundwater bores was assessed to determine whether stream depletion calculations were robust, considered impacts on all nearby streams, were based on reasonable storage and transmissivity values and were transparent in their application. This work provides a critical desktop assessment of the uncertainties surrounding values in some of the larger groundwater takes, the impact of pumping test quality and streambed conductance assumptions. Results are compared to streambed conductance fieldwork completed in the Otago region. Field investigations demonstrated the demonstrated how stream appearance can be highly misleading. The findings from this work emphasise the need to incorporate uncertainty analysis into resource consenting and improve the transparency and consistency of assessments, and also the value of field investigations in sensitive catchments.