IntroductionLed outdoor activities (LOA), defined here as instructed activities taking place in an outdoor setting that have an educational goal associated with them, can involve a degree of uncertainty and risk. Participating in LOA activities that involve an acceptable level of risk can be beneficial as they provide an opportunity for participants to engage with risk, to test their capabilities, and to learn from mistakes (Cline, 2007; Dickson, 2012). Despite LOA organisations' best efforts to keep levels of risk acceptable, the dynamic environment in which many LOAs take place means that LOA leaders and their participants can sometimes find themselves in situations that their organisation has not anticipated, and hence are outside their organisation's procedures or their own experiences or knowledge. These situations may have the potential to negatively impact the safety of the activity leaders and participants. Research from other domains has shown that one means of devising solutions to such unplanned for safety critical situations is through improvisation (e.g. Weick, 1993; Mendonca, 2007; Grotan, Storseth, Ro, & Skjerve, 2008). This study considers safety-related improvisation in LOAs.Improvisation is an ambiguous concept in relation to safety. Inappropriately devised or poorly executed improvisation has contributed to adverse outcomes for activity leaders and participants. Examples include the Mangatepopo Gorge incident in 2008, in which six students and their teacher were swept over a spill weir and drowned after following their LOA leader's improvised plan to exit the flooded Mangatepopo gorge in New Zealand (Brookes, Smith, & Corkill, 2009), and the Ptarmigan Peak incident in Alaska, in which two students died after the failure of the improvised roping and anchoring system devised by the activity leaders led to all four roped climbing groups becoming detached from the slope and falling into a boulder field (Williamson, Ratz & Miller, 1997). To the contrary, evidence from other safety critical domains such as firefighting and emergency response services (e.g. Klein, 1999, pp.19-20; Bigley & Roberts, 2001; Roux-Dufort and Vidiallet, 2003), and aviation (e.g. National Transportation Safety Board, 2010) indicates that appropriate, effective improvisation can save lives. This suggests that improvisation by LOA leaders and their organisations, if appropriate and effective, may also have the potential to impact positively on the safety of participants and leaders in the event that they find themselves in an unanticipated, safety critical situation. An improved understanding of improvisation in relation to LOAs may, therefore, provide new insights into safety for LOA organisations. As a first step in this line of inquiry, research is required to understand the nature of improvisation during LOAs.This article presents the findings from an exploratory study that was undertaken as part of a wider research program examining improvisation by LOA leaders. Specifically, a survey study was undertaken to provide broader evidence that improvisation takes place in safety critical situations within the LOA domain. The aims of this study were to identify the circumstances in which safety-related improvisation occurs, the forms in which it manifests, and to establish what factors influence safety-related improvisation by LOA leaders.What is improvisation?Improvisation, as it is referred to in this study, is an adaptive strategy used when faced with situations for which no procedures exist, or where circumstances prevent known procedures from being deployed. It is not considered as a replacement for well designed emergency procedures. Various definitions of improvisation exist in the literature. For example, it is defined as "creative and spontaneous behaviour of managing an unexpected event" (Magni, Proserpio, Hoegl, & Provera, 2009, p. 1045), and an unplanned response that takes place in a turbulent, rapidly changing environment (Chelariu, Johnston, & Young, 2002). …