Abstract Denitrification beds are a cost-effective technology for removing nitrate from point source discharge. To date, field trials and operational beds have primarily used wood media as the carbon source; however, the use of alternative more labile carbon media could provide for increased removal rate, lower installation costs and reduced bed size. While previous laboratory experiments have investigated the potential of alternative carbon sources, these studies were typically of short duration and small scale and did not necessarily provide reliable information for denitrification bed design purposes. To address this issue, we compared nitrate removal, hydraulic and nutrient leaching characteristics of nine different carbon substrates in 0.2 m 3 barrels, at 14 and 23.5 °C over a 23-month period. Mean nitrate removal rates for the period 10–23 months were 19.8 and 15 g N m −3 d −1 (maize cobs), 7.8 and 10.5 g N m −3 d −1 (green waste), 5.8 and 7.8 g N m −3 d −1 (wheat straw), 3.0 and 4.9 g N m −3 d −1 (softwood), and 3.3 and 4.4 g N m −3 d −1 (hardwood) for the 14 and 23.5 °C treatments, respectively. Maize cobs provided a 3–6.5-fold increase in nitrate removal over wood media, without prohibitive decrease in hydraulic conductivity, but had higher rates of nutrient leaching at start-up. Significant difference in removal rate occurred between the 14 and 23.5 °C treatments, with the mean Q 10 temperature coefficient = 1.6 for all media types in the period 10–23 months.