Lake Okaro is a small, warm monomictic lake in the central North Island of New Zealand. It has remained highly eutrophic despite an intensive catchment and in-lake restoration program which commenced in 2003. The program has included the implementation of a constructed wetland, riparian protection, an alum application and application of a modified zeolite mineral (Z2G1) to reduce internal nutrient loading. This study examines water column and sediment nutrient dynamics; focusing on phosphorus (P) and the ecosystem response to lake restoration designed to reduce levels of P. Trends in P concentrations in Lake Okaro were linked to the restoration efforts over a six-year period (2002-08) including the period shortly before the restoration program. Over the entire study period, the annual average total phosphorus (TP) concentration in the lake decreased by 56 %. Two predictive models, which derive the annual average P concentration in the water column based on external P loading, generally underestimated the measured TP concentrations in the water column due to internal P loading. Of all restoration methods, the application of Z2G1 produced the most effective reduction in water column TP concentrations. However the lake trophic state showed high resilience to reduced internal P loading even though the combined effect of all restoration procedures resulted in significantly decreased TP concentrations in the lake. The sources and sinks of nutrients in the hypolimnion of Lake Okaro were investigated using field measurements in a comprehensive nutrient budget model in order to determine changes in sediment nutrient fluxes resulting from a whole lake sediment capping trial using Z2G1. Sediment nutrient fluxes in the hypolimnion were estimated as the residual term in the nutrient budget model that accounted for mineralisation of organic nutrients, nutrient uptake by phytoplankton, nitrification, adsorption or desorption of P from inorganic particulate material in the water column, and diffusion of dissolved nutrients at the thermocline. The model indicated that during a period of seasonal stratification in 2007-08 up to 60% of hypolimnetic phosphate fluxes and 50% of ammonium fluxes were derived from bottom sediments. Diffusion across the thermocline, adsorption/desorption of phosphate to suspended solids, and nitrification were of relatively minor importance (less than or equal to 9%) to the total fluxes. Any reduction in sediment nutrient release by Z2G1 was small compared with both the total sediment nutrient flux and the sum of other hypolimnetic fluxes. Sediment and settling seston organic P composition was determined using 31P nuclear magnetic resonance (NMR). Settling seston and sediment samples were analysed during winter and summer, representing, respectively, a mixing period when the water column was well oxygenated and a stratified period when the hypolimnion was anoxic. The bottom sediments and settling seston contained orthophosphate, orthophosphate mono- and diesters, pyrophosphates, polyphosphates, and phosphonates with organic P content exceeding 60% of the total extracted P occasionally. Phosphorus content in settling seston increased 2.5-fold in winter, with a marked increase in orthophosphate content. The 31P NMR analyses revealed the presence of several potentially bioavailable P compounds, which may be recycled from the sediment to the water column. An apparent half-life‟ value was used to quantify the time scales on which these compounds are degraded within the sediment and likely being recycled to the overlying water column. Relatively long half-life values, ranging from 8 to 23 years, indicate that this recycling could potentially reduce the efficacy and longevity of in-lake restoration procedures that have been applied to Lake Okaro. A one-dimensional process based ecosystem model (DYRESM-CAEDYM) was used to simulate the potential effect on water quality of Lake Okaro of separate and combined reductions in external and internal loads of nitrogen (N) and P. The model was calibrated against field data for a two-year period and validated over two separate one-year periods including a year immediately following a Z2G1 application and a year when there was an extraordinary algal bloom from an invasive, highly buoyant, N-fixing cyanobacterium, Anabaena planktonica. The model simulations reproduced the scale of phosphate and ammonium concentrations at 14 m depth, corresponding to the deeper region of the hypolimnion, both before and after the application of Z2G1, with no adjustment of parameters, suggesting that there was little effect of the Z2G1, at least within the uncertainties of the model runs. The model simulations were less successful in reproducing the Anabaena planktonica bloom. This was attributed to a lake of flexibility in the conceptualisation and calibration of the model, which meant that it could not encompass this invasive species. In the model scenarios with reduced nutrient loading, the trophic status of Lake Okaro, given quantitatively by the Trophic Level Index (TLI), decreased to a greater extent with a given fractional reduction of the internal load than a reduction of the external load. The control of both N and P was shown in simulations to be more effective in reducing phytoplankton biomass than for N or P alone, tending to affirm an N+P control paradigm. Undesirable shifts in zooplankton and phytoplankton species composition due to the application of Z2G1 were investigated by comparing the plankton community structure before and after the Z2G1 application. No significant differences in species composition were found at the depths investigated (surface and 9 m). However, further analyses showed statistically significant differences between seasons, indicating that seasonal variations in plankton composition far outweighed changes that occurred as a result of the Z2G1 application. In this study, field measurements and numerical modelling provided a comprehensive assessment methodology of testing the response of Lake Okaro to reduced nutrient loading, with a focus on P dynamics. Although nutrient loads to Lake Okaro were reduced using catchment and in-lake restoration methods, further substantial and prolonged reduction in both N and P loading appear to be required to decrease phytoplankton biomass and the trophic state of the lake. This study highlights the need for investigating water column P content and sediment P composition for evaluating the potential magnitude of internal loading, and emphasises the importance of using process based numerical modelling as a decision support tool for lake management.