A predictive model based on the theory of phase space reconstruction was proposed to study the determinism and predictability of dynamics underlying pressure fluctuations by measuring and analyzing the time series of pressure signals at different locations in a bubble bed with 0.3 m in dia. and 3 m in height. Chaotic invariants (correlation dimension, K2 entropy, and Lyapunov exponent spectrum) of measured and model-generated time series of pressure signals were nearly the same. The model captured some important nonlinear characteristics of the real system, which can be used to study the dynamics of fluidizing system. Deterministic dynamics underlying pressure signals were confirmed to exist. A new characteristic index defined as the determining level of dynamics was used to analyze deterministic degrees at different gas velocities and locations of the wall in the bed. Since prediction errors of pressure fluctuations grow exponentially with time at short time scales and the exponent separation rate between predicted and measured values is proportional to the maximal Lyapunov exponent, the long-term predictability of pressure fluctuations is impossible. This verified chaotic properties of fluctuation dynamics in the fluidized bed with deterministic mechanism and sensitive dependence on initial conditions.