Economic mass production is a crucial stage in the development of biocontrol agents. This paper compares different solid substrates with and without preceding liquid culture for efficacy and economics in the scaled-up production of the mycoparasitic biocontrol agent Clonostachys rosea. In a basic liquid medium (BLM) containing (in g l−1): molasses (80), neopeptone (10) and yeast extract (2), Clonostachys rosea produced 107 conidia ml−1 after 4 d in agitated culture. Spore viability was 99% and did not differ from that of conidia derived from potato dextrose agar plates. Guata, an inert polyester fibre, coated with BLM and cassava starch (30 g l−1), was the best solid matrix yielding 109 conidia g−1 after 20 d. On a volume basis after extraction and on a production unit basis, Guata was in a similar range to Biodac®, a granular cellulose complex, which had also been enriched with BLM, producing up to 107 conidia ml−1 or over 1010 conidia per 23×28 cm polypropylene bag, the production unit in which all solid matrices were incubated. Rice was the lowest-yielding solid matrix regardless of quality. Chlamydospores were not observed in liquid or on solid media. All conidia produced on solid matrices had a viability >90%. Conidia produced on Biodac had a higher viability (95.6%) than spores produced on Guata (93.1%) after 24 h on potato-dextrose plates. However, production on Biodac was more variable and negatively affected by prefermentation in liquid culture, whereas production on Guata and rice was unaffected by the origin of the starting inoculum. Guata of medium thickness (1.27 cm) was superior to thicker (2.54 cm) or thinner (0.64 cm) sheets with respect to both yield per gram and per production unit. Although prefermentation neither improved yield nor reduced cost, it increased throughput by a factor five and may be advantageous where fermentor and/or incubator space is limited. Biodac and Guata were the most economical solid matrices; inoculum to treat one hectare (3×1012 conidia) cost US $ 27, whereas rice-based production cost US $ 613 under Costa Rican conditions. Liquid culture alone would be fast but too costly: US $ 1078 ha−1. Experiments to further reduce costs in the Guata system are suggested.