The ability of groundwater models to inform recharge through calibration is hampered by the correlation between recharge and aquifer parameters such as hydraulic conductivity (K), and the insufficient information content of observation datasets. These factors collectively result in non-uniqueness of parameter estimates. Previous studies that jointly estimate spatially distributed recharge and hydraulic parameters are limited to synthetic test cases and/or do not evaluate the effect of non-uniqueness. The extent to which recharge can be informed by calibration is largely unknown for practical situations, in which complexities such as parameter heterogeneities are inherent. In this study, a systematic investigation of recharge, inferred through model calibration, is undertaken using a series of numerical experiments that include varying degrees of hydraulic parameter information. The analysis involves the use of a synthetic reality, based on a regional-scale, highly parameterised, steady-state groundwater model of Uley South Basin, South Australia. Parameter identifiability is assessed to evaluate the ability of parameters to be estimated uniquely. Results show that a reasonable inference of recharge (average recharge error <10%) requires a surprisingly large number of preferred value regularisation constraints (>100 K values across the 129 km2 study area). The introduction of pumping data into the calibration reduces error in both the average recharge and its spatial variability, whereas submarine groundwater discharge (as a calibration target) reduces average recharge error only. Nonetheless, the estimation of steady-state recharge through inverse modelling may be impractical for real-world settings, limited by the need for unrealistic amounts of hydraulic parameter and groundwater level data. This study provides a useful benchmark for evaluating the extent to which field-scale groundwater models can be used to inform recharge subject to practical data-availability limitations.