Application of Probabilistic Risk Assessment (PRA) and Deterministic Risk Assessment (DRA) at a coking plant site was compared. By DRA, Hazard Quotient (HQ) following exposure to Naphthalene (Nap) and Incremental Life Cancer Risk (ILCR) following exposure to Benzo(a)pyrene (Bap) were 1.87 and 2.12 × 10^-4. PRA revealed valuable information regarding the possible distribution of risk, and risk estimates of DRA located at the 99.59th and 99.76th percentiles in the risk outputs of PRA, which indicated that DRA overestimated the risk. Cleanup levels corresponding acceptable HQ level of 1 and ILCR level of 10^-6 were also calculated for both DRA and PRA. Nap and Bap cleanup levels were 192.85 and 0.14 mg·kg^-1 by DRA, which would result in only 0.25% and 0.06% of the exposed population to have a risk higher than the acceptable risk, according to the outputs of PRA. The application of PRA on cleanup levels derivation would lift the cleanup levels 1.9 times for Nap and 2.4 times for Bap than which derived by DRA. For this coking plant site, the remediation scale and cost will be reduced in a large portion once the method of PRA is used. Sensitivity analysis was done by calculating the contribution to variance for each exposure parameter and it was found that contaminant concentration in the soil (C_s), exposure duration (ED), total hours spent outdoor per day (ET_out), soil ingestion rate (IR_s), the air breathing rate (IR_a) and bodyweight (BW) were the most important parameters for risk and cleanup levels calculations.