By neglecting the natural variability of structural parameters and walking loading, design guidelines featuring deterministic approach to calculating pedestrian-induced vibrations of long-span floors do not provide a rational prediction of the actual vibration response. Considering uncertainties of input parameters and assessing the probability of exceeding vibration limits lead to a more informative and reliable evaluation of floor vibration performance. This study presents an uncertainty analysis of modal properties and pedestrian-induced acceleration response of a multi-panel CLT floor, taking into account variability in timber material properties, inter-panel connections and damping ratio. A sensitivity analysis revealed that timber density, longitudinal elasticity modulus and shear modulus had the greatest impact on the natural frequencies and modal masses. In contrast, variability in the inter-panel connections had only a minor influence. The uncertainty analysis of the vibration response was carried out using Monte Carlo simulation, where the timber material properties and a damping level were uncertain. The pedestrian loading was modelled deterministically, accoriding to a popular design guideline for floor vibrations. Two sets of vibration responses were simulated numerically: (i) with a fixed walking frequency across all simulations, and (ii) with variable walking frequency adjusted to induce a resonance in each simulation. The resonant response also was calculated following the guideline for comparison. The first dataset showed that the guideline's response had a very low probability of occurrence. On the other hand, it compares reasonably well with the second dataset, indicating that the guideline provides rational vibration levels only for a likely occurrence of resonance.