Traditional risk assessment and source apportionment of sediments based on bulk polycyclic aromatic hydrocarbons (PAHs) can introduce biases due to unknown aging effects in various sediments. We used a mild solvent (hydroxypropyl-β-cyclodextrin) to extract the bioavailable fraction of PAHs (a-PAHs) from sediment samples collected in Pearl River, southern China. We investigated the potential application of this technique for ecological risk assessments and source apportionment. We found that the distribution of PAHs was associated with human activities and that the a-PAHs accounted for a wide range (4.7%–21.2%) of total-PAHs (t-PAHs), and high risk sites were associated with lower t-PAHs but higher a-PAHs. The correlation between a-PAHs and the sediment toxicity assessed using tubificid worms (r = −0.654, P = 0.021) was greater than that from t-PAH-based risk assessment (r = −0.230, P = 0.472). Moreover, the insignificant correlation between a-PAH content and mPEC-Q of low molecular weight PAHs implied the potiential bias of t-PAH-based risk assessment. The source apportionment from mild extracted fractions was consistent across different indicators and was in accordance with typical pollution sources. Our results suggested that mild extraction-based approaches reduce the potential error from aging effects because the mild extracted PAHs provide a more direct indicator of bioavailability and fresher fractions in sediments.