Several models predict soil pore space indices (the relative gas diffusion coefficient, D_s/D_o and the pore tortuosity, t), but information is lacking on which models predicted indices better relate to soil processes. We compared pore space indices′ predictive models based on air-filled porosity (f_a) alone vs. models using air-filled porosity and total pore space (Φ) (f_a Φ). We also assessed the relationships between these indices and CO₂ and N₂O. The study was conducted from 2011 to 2014 on a silt loam soil at Lincoln University. Soil samples were collected at 0-10 and 10-20 cm depth and oven dried at 105 °C for 72 h. After drying, f_a and Φ were calculated and later used in models for predicting D_s/D_o and t. CO₂ and N₂O were measured with a Shimadzu gas chromatograph (GC) and a photoacoustic gas analyzer (PSA). Results showed that D_s/D_o predicted using f_a alone (Marshall and Buckingham) was higher as compared with values predicted with models based on f_a Φ (Sallam et al., Millington, and Jin and Jury) (P < 0.001). However, values of t predicted with models based on f_a alone were lowest (P < 0.001). CO₂ and N₂O measured with GC better related with D_s/D_o and t.