The first National Recreational and Indigenous Fishing Survey in 2000-01 was a significant breakthrough, for the first time providing a comprehensive national picture of participation, effort, catch, and expenditure. The standard sampling methodology developed at the time was based on a randomised survey of the Australian telephone directory. Due to declining land line use this sampling frame has become non-representative of the Australian population, and rapid change in availability of alternative sampling frames highlighted a need to explore alternative sampling approaches. The National Recreational Fishing Survey has investigated the use of blended sampling that recruits people using more than one sample frame and sampling method in order to ensure coverage of all types of recreational fishers. This included exploring the use of non-probability based sample recruitment methods, with a focus on ensuring random selection and high quality design of sampling when using non-probabilistic approaches. The survey trialled multiple methods of recruitment and assessed whether model-based weighting could be used to generate robust estimates of social and economic aspects of recreational fishing. The outcomes offer new insights into recreational fishing survey design for the future.