The Swiss registry of charges concerning illegal trade and consumption of narcotics makes a distinction between first and subsequent charges as well as between several types of illegal drugs. It has been previously shown by using a mathematical model that changes in incidence of illegal drug use can be inferred from the time series of two indicators that are calculated from the numbers of first and subsequent charges. In the current study the method of data analysis is refined in order to monitor the incidence of heroin/cocaine use. The drug career of the great majority of heroin/cocaine users begins with a stage in which they use cannabis and legal drugs only. There are two possibilities of modeling charges against drug users with two stages of illegal drug-using behavior. The first model has four compartments that correspond to the charged and the no-yet- charged users at each of the two stages. This model cannot differentiate between sojourn times at stage 1 of persons who leave stage 1 and become drug-free and those who proceed to stage 2. The second or extended model has six compartments, with the two additional compartments representing those who are "at risk" for proceeding to stage 2. This extension offers the possibility of choosing the duration of drug use of non-proceeders independently from the sojourn time at stage 1 of those who proceed to stage 2. This is an advantage when there is a subpopulation of long-time drug users who do not proceed to stage 2. This study concludes that 2,500 persons per year may have entered the population of habitual heroin/cocaine users in Switzerland in the early 1980's, and approximately 10,000 each year in 1991-93. After 1993, however, the incidence of heroin/cocaine use declined rapidly. Sociological explanations of these trends are offered. 6 figures, 3 tables, and 4 references