While most research on particle tracking velocimetry (PTV) is devoted either to 2D flows or to small-scale 3D flows, this paper describes a complete 3D PTV algorithm and some applications to indoor airflow velocity measurements. A particle detection procedure especially adapted to the physical characteristics of neutrally buoyant helium-filled soap bubbles is described. To recover longer trajectories, a temporal tracking algorithm is used based on polynomial regression and including a cross-gap and a forward-backward strategy. In order to increase the measurement area and the number of trajectories, the correspondence problem is addressed by a new procedure involving fundamental matrices from both a three and a two-camera arrangement. 3D reconstruction is done by a least-squares method. Some guidelines are given in terms of camera and light positioning for 3D PTV in large volumes. Applications of the algorithm include Lagrangian tracking in (i) a 3.1 m × 3.1 m × 2.5 m light-gray walled test-room; (ii) a 5.5 m × 3.7 m × 2.4 m black walled test-room; (iii) over a heat source; (iv) inside an experimental aircraft cabin. Results show that the algorithm is capable of tracking more than 1400 tracers in volumes up to 3 m × 3 m × 1.2 m.