This article describes a probabilistic site‐response analysis for the city of Oslo, Norway. We first perform a probabilistic seismic hazard analysis (PSHA) for hard rock. Then, we conduct site‐response analyses using Monte Carlo simulations to capture uncertainty in the site profile. We include four base‐case soil profiles to incorporate epistemic uncertainty, and we vary the shear‐wave velocity profile and shear‐modulus reduction and damping curves to account for aleatory variability. We base the soil profiles on over 7000 in situ tests, and the shear‐wave velocity profile median, standard deviation, and interlayer correlation on over 559 cone penetration tests. Next, we perform regression analyses to estimate medians and standard deviations of site‐specific amplification factors (AFs). Finally, we modify the ground‐motion models for rock with the AFs and recompute the PSHA for the soil surface. The analyses show that (1) shallower soil profiles have larger uniform hazard spectra (UHS) values at short periods and smaller UHS values at long periods; (2) epistemic uncertainty of the base‐case soil shear‐wave velocity profile leads to alternative UHS values with a difference of a factor of 2 at short periods; (3) there is only a small difference in the mean magnitudes and distances controlling the hazard for the PSHA conducted for rock compared to soil; (4) response spectra calculated from site‐response analyses with no aleatory variability of the soil properties predict significantly smaller spectral acceleration values at periods shorter than the natural site period; (5) using site amplification standard deviations based on ground‐motion recordings, instead of site‐response analyses, results in a 10%–20% reduction in the soil surface UHS at short periods and a 5%–10% increase at long periods; and (6) the Eurocode 8 AFs are, in general, conservative for Oslo.