Freshwater floodplain wetlands in Queensland’s Great Barrier Reef Catchment Area are surveyed annually as part of a long-term monitoring program to model and report on trend in condition, based on a suite of pressure and state indicators. Logistical constraints and the voluntary nature of wetland inclusion means some wetlands originally earmarked for surveying cannot be surveyed, which creates complexity in the spatially balanced, serially alternating panel design used for sampling. Any ‘non-responding’ wetlands are replaced by other randomly selected wetlands that can be surveyed (i.e., ‘responding’ wetlands). This results in a sample representing responding wetlands only, rather than all wetlands in the target subpopulation, and must be accounted for to produce unbiased estimates of trend. By way of an example, we outline statistical methods to address the non-response problem, including decisions that need to be made concerning the nature of the non-response (can it be considered as missing at random, or otherwise?) and the missing-data distributions (are they unknown, or can assumptions be made?). The methods and example we outline will be of interest to those designing, implementing or analysing data from freshwater monitoring programs, particularly when planned and unplanned complexity of survey design are involved.