To reduce the field effort required for 2-D and 3-D shallow seismic surveying, we have developed a towed land‐streamer system. It was constructed with self‐orienting gimbal‐mounted geophones housed in heavy (1 kg) cylindrical casings, sturdy seismic cables with reinforced kevlar sheathing, robust waterproof connectors, and a reinforced rubber sheet that helped prevent cable snagging, maintained geophone alignment, and provided additional hold‐down weight for the geophones. Each cable had takeouts for 12 geophones at 1 m or 2 m intervals. By eliminating the need for manual geophone planting and cable laying, acquisition of 2-D profiles with this system proved to be 50–100% faster with 30–40% fewer personnel than conventional procedures. Costs of the land‐streamer system and total weight to be pulled could be minimized by employing nonuniform receiver configurations. Short receiver intervals (e.g., 1 m) at near offsets were necessary for identifying and mapping shallow (<50 m) reflections, whereas larger receiver intervals (e.g., 2 m) at far offsets were sufficient for imaging deeper (>50 m) reflections and estimating velocity‐depth functions. Our land‐streamer system has been tested successfully on a variety of recording surfaces (e.g., meadow, asphalt road, and compact gravel track). The heavy weight of the geophone casings and rubber sheet ensured good geophone‐to‐ground coupling. On the asphalt surface, a greater proportion of high‐frequency (above 300–350 Hz) energy was recorded by the land streamer than by standard baseplate‐mounted geophones. The land‐streamer system is a practical and efficient means for surveying in urbanized areas. Acquisition and processing of 3-D shallow seismic data with the land‐streamer system was simulated by appropriately decimating and reprocessing an existing 3-D shallow seismic data set. Average subsurface coverage of the original data was ∼50 fold, whereas that of the simulated data was ∼5 fold. The effort required to collect the simulated pseudo-3-D data set would have been approximately 7% of that needed for the original field campaign. Application of important data‐dependent processing procedures (e.g., refraction static corrections and velocity analyses) to the simulated data set produced surprisingly good results. Because receiver spacing along simulated cross‐lines (6 m) was double that along in‐lines (3 m), a pattern of high and low amplitudes was observed on cross‐sections and time slices at early traveltimes (⩽50 ms). At greater traveltimes, all major reflections could be identified and mapped on the land‐streamer data set. With this cost‐effective approach to pseudo-3-D seismic data acquisition, it is expected that shallow 3-D seismic reflection surveying will become attractive for a broader range of engineering and environmental applications.