The data used in the study are from the Travel and Tourism database (TDB) and cover the years 1990-1996. The TDB contains information on individual/household travel behaviour in Sweden (both business and leisure) and are obtained by monthly telephone interviews of 2000 randomly sampled Swedish citizens aged 0-74 years. The survey contains, among other things, information on the number of overnight trips made during the previous month along with socio-economic information. For the two latest trips, detailed information on, e.g., origin and destination of the trip, main purpose of the trip, number of days spent at main destination, mode of transportation, and expenditure at destination are available. A drawback with TDB is that it does not report transportation costs for the trips. The present study is focused on Swedish households' leisure travel behaviour and therefore business trips are excluded from the sample. Further, only overnight trips to the three largest city regions, Stockholm, Gothenburg and Malmö, are considered. In the final version of the paper, only results for the Stockholm sample are reported. The sample is selected so that only households with an annual income less than 1 million Swedish krona (SEK) and households whose oldest individual is between 18-65 years of age are considered. Households reporting more than 30 nights in total for the month were also excluded from the sample. The sample is further restricted to include only households that recorded two or less trips in total regardless of the destination (consult paper for explanation). The total number of households in the sample (prior to enforcing restrictions) are 185 301 of which 28.2 percent made at least one trip to any destination. Out of these, 33 percent or 17 522 households travelled at least once to one of the major city regions considered in the paper. The reported Stockholm sample (after enforcing the restrictions) consist of 6401 observations for the second part of the hurdle model and 13 531 observations for the first part of the hurdle model. The following variables from the TDB are used directly, or used in calculation, of the desired variable (definitions from the TDB): Sfo Number of overnight trips f1totov + f2totov Total number of nights spent at the trips f1ptotal, f2ptotal Cost on location inkom Income for individual being interviewed f1syfte1 Purpose dummy (1= visiting relatives and friends) datum July dummy palder1:19 number of children at different ages orts Dummy for origin of the trip palder1:19 Age of oldest family member on the trip sysselb to calculate leisure time*. *Full time workers and individuals in military service are assumed to have 220 working days per year. Part time workers and individuals working at home are assumed to have 110 working days per year. Students are assumed to possess more flexibility concerning their work time and are assumed to possess 85 work days per year. Unemployed are assumed to have 365 leisure days per year. The absolute values of the leisure time variable may be questioned, but hopefully it captures increasing levels of flexibility concerning leisure time. Hence, full time workers and individuals in military service are assumed to have 145 leisure days, part time workers and individuals working at home 255, students 280, and the unemployed 365 leisure days during a year. Dummies are constructed for each category to catch any misspecification. The following variables have been calculated partly using TDB and other information: Cost of transportation: For car travel, travelled distance (based on origin, f1stort and f2stort, and destination f1hort and f2hort obtained from the TDB and a distance matrix, supplied in both the SPSS file avstand.sav and the ASCII file avstand.dat) is used to compute the cost. It is assumed that decision makers only consider direct costs, i.e. gas and maintenance. The average monthly gas prices during each year 1990-1996 are used (supplied in an SPSS syntax file, Gas.sps). Data on fuel consumption and other operating costs are obtained from the Swedish Automobile Association. Motor, 1990-1996, M.s Bilkostnadskalkyl, Stockholm. The travel time for cars assumes an average speed of 75 km per hour, and an additional 30 minutes rest for every four hours of driving. Bus costs are calculated using a ticket price per km obtained from bus-price tables (supplied in a SPSS syntax file, bussp.sps) and the travel times for bus are derived based on an average speed of 60 km per hour. Travel times for bus includes 45 minutes rest every four hours. For air transportation, costs and travel times are calculated using price and time tables obtained from SAS (Scandinavian Airline Systems, supplied in a SPSS syntax file, flyg.sps). Train costs (supplied in a SPSS syntax file, tagpris.sav) are calculated using an average ticket price per km obtained from the Swedish Railway. It is assumed that travellers that travel more than 600 km buy a sleeper ticket with a price corresponding to an average of the price of sleeper tickets in compartments with three beds and six beds. Train travel times are obtained from train time tables (supplied in a SPSS syntax file, tagtid.sps). Summary descriptions of the samples and details concerning calculations of explanatory variables are given in a previous version of the paper available at www.econ.umu.se/~jorgen.hellstrom