This paper contributes to a statistical debate about the measurement of the size and growth of the Tanzanian manufacturing sector in the last decade. Evidence from official data sources shows a contraction of industrial output in the early 1990’s but sustained expansion since 1995. There is a problem however in the coverage of the quarterly survey used to calculate these growth rates, which only includes a limited selection of firms with 50 employees or more. Evidence from panel survey data, which covers a more limited range of industries but all size categories, indicates a continuing decline in aggregate real output and employment levels since 1995. We present a number of descriptive statistics which seek to identify the potential sources of these differing pictures of trends in industrial growth. We show that large firms in our sample have performed differently from small firms, showing a sharp contraction in both real output and employment before 1995 and some signs of recovery from 1995-98, which is consistent with the trend observed in the official data. However, small firms (those with < 50 employees) have actually contracted more sharply since 1995. Hence, by ignoring these sustained declines in the small firm sector, official statistics are presenting an artificial picture of a rebound in the fortunes of the Tanzanian manufacturing sector since 1995. We discuss some possible explanations for the different responses of large and small firms to changing policy and business operating conditions during this period. These findings have relevance for Tanzanian policy-makers and researchers seeking to design policy towards the industrial sector.