The effect of turbulence closure on spray dynamics is studied for three dilute Acetone spray jets using an Eulerian–Lagrangian approach with two-way coupling. A stochastic random walk algorithm is employed to model the droplets’ dispersion. Simulations using different turbulence closure models (i.e. the isotropic SST-k-ω, and the k-∊ realizable models, and the non-isotropic Reynolds Stress Model (RSM)) are compared with the Sydney spray measurements for SP2,SP6, and SP7 spray conditions (Gounder et al., 2012). The experimental mass flow rate, spray mean and rms data are injected for each available size bin during the numerical simulations. Overall, the simulations show good comparisons with the mean and rms spray measurements. The turbulence closure non-isotropy modeling shows relatively weak effect on the spray mean velocity and size profiles predictions, where the RSM predictions was slightly better at the centerline at x/D=10 and x/D=20 for the SP2 mean axial velocity, with similar predictions to the k models for the SP6 and SP7. The RSM, however, consistently predicts better rms axial and radial spray velocity distribution, which indicates more realistic droplet dispersion than the isotropic SST-k-ω and the k-∊. The RSM gas phase shear stresses show that close to the nozzle (i.e. at x/D=5 and x/D=10) the turbulence non-isotropy increases as we approach the centerline and is maximum at the shear layer location at x/D=5. On the other hand, at the downstream locations at x/D=30 the turbulence field non-isotropy increases at the jet edge. A conical region of large size mean droplets (D10>26μm) is observed around the SP2 jet edge. This region disappears quickly for the SP6 and SP7 after the inflow section. The data analysis exhibits that the RSM predicts higher Stokes number than the SST-k-ω due to faster mixing time scales. For the impact on dynamics, at the centerline the SST-k-ω and k-∊ models, predict higher droplets’ slip velocity, higher drag force, and faster droplet’s response than the RSM. The droplets’ tendency for cross-stream dispersion is also found to vary with the turbulence model. The fan spreading phenomenon is observed at the down stream locations away from the nozzle. The results show that the spray turbulence is non-isotropic and lags the gas phase rms values, especially at the centerline. This discrepancy decreases downstream and towards the jet edge. The study shows the importance of non-isotropy modeling on the droplets dispersion and spray dynamics predictions.