The relationship between smoking parameters and odour characteristics, evaluated by a trained sensory panel.. were studied on smoked herring. In addition, a possible correlation between the content of 10 phenolic compounds and sensory perceptions was investigated. Five smoking techniques were applied, combining smoke production conditions, performed by pyrolysis of beech wood sawdust or by friction of beech wood log, with smoke deposition, either in a controlled kiln (traditional smoking) or by an electrostatic process. In the fifth smoking technique, a purified condensate of beech smokes was vaporised on fish fillets in the smokehouse. The time of smoking was 3 It for traditional smoking and the liquid smoke atomisation process and 12 min for the electrostatic method. The effects of three smoking temperatures (16, 24, 32 degrees C) were tested for both the traditional and the liquid smoke atomisation processes, as well as the effect of the position of the exhaust valve in the smokehouse in the case of the traditional method. Two different voltages were applied for the electrostatic process, 37 and 42 kV. The results show a clear discrimination of the products since some odour characteristics are specifically related to the smoking process applied. All the studied parameters (smoke generation, deposition of smoked compounds, smoking temperature, exhaust valve opening in the smokehouse or voltage applied in the electrostatic tunnel) have an effect on the smell characteristics of smoke products, either on the odour intensity and/or on the kind of smoke note. Multiple linear models were tested to find relationships between sensory properties and phenolic compounds. Although some compounds seem to be mainly involved in the "cold ash" note, the results illustrate the difficulty of reaching clear conclusions about a correlation between smoke odour and only 10 phenolic compounds. It is suggested that a better model could be found if other volatile compounds, besides the phenolic class, are taken into account.