Results and policy conclusions: Firstly, several factors are significant for positive employment impacts. With respect to the reference category (logistics innovations), product and service innovations have a positive effect, while all other innovation categories are not significant. This is in line with the general literature on innovation and employment. Indirect substitution effects of product and service innovations on the firm level appeared to be lower than we expected. To control for this substitution effect, the question. Did this innovation replace some of the product/service sales of your establishment? was used. While we expected substitution effects in nearly all cases, only 43% of the product innovators and 27% of the service innovators answered yes. It seems that ecological products and services have created their own market niches being supplements to conventional goods and services. However, it can be assumed that for most of the innovations mentioned by the firms, as for example ¿new cleaning techniques or transport reduction measures, negative indirect substitution effects occur in other firms. Small firms report more employment increases than large firms. This result is in line with other empirical studies on the general relation between innovation and employment, too. Firms with high shares of employees with college or university degrees have a higher probability to increase employment in the wake of innovations. This may be an indication that also environmentally-oriented innovations are skill-biased. Positive sales expectations are highly significant for increasing employment. Firms in Germany, the Netherlands and Italy have a higher probability to report a positive employment effect than firms in the UK, the reference country. Secondly, some other factors have been identified correlating significantly with an increasing probability of job losses. End-of-pipe-processes have positive, recycling innovations a negative correlation with the probability that the innovation has a negative impact on employment. This result is quite surprising since we expected the same direction of employment effects for both kinds of innovations. However, already the descriptive analysis has shown that employment effects of recycling innovations are positive in almost all cases. An explanation for the difference between end-of-pipe technologies and recycling measures may be the maturity of regulation. While end-of-pipe regulations have existed, in many cases, for twenty or thirty years, political measures concerning recycling have been mainly implemented during the nineties. Thus, new end-of-pipe innovations often replace existing older technologies, while other environmental process innovations, especially in the area of recycling, have lower substitution effects. We have addressed this substitution effect by the following two questions in the questionnaire "Did the introduction of this process innovation replace, at least in part, a previous production process in your establishment? and "Did the introduction of this process innovation replace, at least in part, end-of-pipe pollution control equipment? 38% of end-of-pipe innovators answered the first question with yes, compared to 19% of recycling innovators. And 34% of end of- pipe innovators agreed on the second question, compared to 13% of recycling innovators. These numbers indicate that substitution effects are substantially different across different types of environmental process innovations. This may explain the better performance of recycling innovations in terms of employment. When environmental goals motivate the innovation, it is less probable that the innovation has a negative employment effect. On the other hand, cost reduction as motivation for the innovation increases the probability that the firm reduces employment. In contrast, innovations aiming at cost reduction have only negative effects on employment. It can be expected that firms invest in improved laboursaving technologies, especially in areas where the technologies have already reached a certain maturity (mainly end-of-pipe technologies). If the firm is not under the pressure of strong cost competition, ¿soft¿ factors like environmental aspects become more important. This decreases the probability of job losses. Further, three factors have been identified which can affect employment in both directions. Market share as an innovation goal, innovation size and strictness of environmental regulation are significant for either positive or negative employment changes. Market-oriented strategies focusing on the development of environmentally beneficial innovations impose chances but also risks on firms. Environmental products and services are often marketed on small niches and thus bear risks of profitability. Other innovation goals like complying with environmental regulations still dominate. Concerning innovation size, the result confirms the hypothesis that fixed turnover costs lead to changes in employment only if there is a major re-organization in the wake of the innovation. A critical innovation size must be reached before employment changes are measurable in person/years by surveys as carried out within our project. Also environmental regulation can lead to both positive or negative employment effects, depending on the concrete innovation activities being undertaken due to the regulatory pressure. Finally, it is remarkable that subsidies or grants for the innovation do not have any employment impact. This means that the allocation of subsidies for environmental technologies and innovations in the respective countries is neutral concerning employment and thus does not counteract labour market policy. However, employment could be stimulated by shifting more money from end-of-pipe measures to integrated measures, especially products and services. But even an employment-oriented allocation of subsidies for environmental programmes will only induce minor employment changes. To sum up the results of the econometric analysis, basic hypotheses about the general relationship between innovation and employment have been confirmed. Product and service innovations create more jobs than process innovations. Employment changes only occur in the wake of major innovations. This is confirmed by the significant impact of the share of innovations expenditures variable for employment increases as well as employment decreases. We also detect signs of skill-biased technological change of eco-innovations, because the share of highly qualified employees has a positive impact on employment increases, while it is insignificant for employment decreases. Significant impacts of the control variable size show that employment is especially created in small firms. A positive correlation between sales expectations and employment is in broad conformity with the innovation literature. Beyond these general insights, some specific conclusions can be drawn for a coordinated environmental, innovation and labour market policy. Generally, eco-innovations have a small but positive effect on employment on the firm level. Thus environmental support programmes do not counteract labour market policy. However it should not be expected that ecological modernisation of industries gives substantial contributions to overcome mass unemployment. A further shift from end-of-pipe technologies to cleaner production, especially towards product and service innovations, would be beneficial for the environment and would create jobs. This synergy should be considered in political programmes whether they are borne by environmental, labour market or innovation policy. Some potential still exists for shifting subsidies from end-of-pipe solutions to cleaner products and services.