It is useful to check index usage at run-time, while XREF is reporting indexes at compile time. It also will give information about query bracketing. Before V9 parameter works only with local clients. In Windows it produces General Protection Faults (GPFs) with Progress 8.1 and below, though it works in 8.3 and above. It does not work with find and can-find. Writes index usage report to a database log file. And it does that for every query, so your log file will grow very very fast. Examples: 1. For each customer where cust-num > 10 It will write to log file 18:23:08 INDEX 12 1 0 (6157) Here '12' is _index._idx-num of the "cust-num" index. 1 0 means one range match. 2. For each customer where cust-num < 10 18:23:08 INDEX 12 0 1 (6157) 3. For each customer where cust-num = 10 18:30:18 INDEX 12 1 1 EQUALITY (6154) That means we have one equality match. EQUALITY word means that all components of the index participate in an eqality match. 4. Let's try another index For each customer where country = "USA" or For each customer where country begins "U" or For each customer where country > "A" and country < "Z" All the above will produce 18:26:40 INDEX 14 1 1 (6157) Here '14' is _index._idx-num of the "country-post" index. We have equality or 2 range matches on the first component of the index. BEGINS is the same as 2 range matches. There is no EQUALITY word here because it is multi-component index and bracket on the first component. In all above examples index had a bracket on the first component only. 5. For each customer where country = "USA" and post > 19115 18:26:40 INDEX 14 2 1 (6157) We have equality on the first component and a range on the second. 6. For each customer where country begins "U" and post > 19115 18:26:40 INDEX 14 1 1 (6157) With "begins" instead of equality on a country field in the same example we have no bracketing on the second component. 7. Finally, For each customer where country = "USA" and post = 19115 will give us 18:30:18 INDEX 14 2 2 EQUALITY (6154) 8. When you have INDEX idx-num 0 0 there are no bracketing, that is the case when you will have WHOLE-INDEX in XREF. In short, the maximum of that two numbers after _idx-num will give us a number of components that we have bracketing on. When a query uses a word index the -zqil parameter is the only way to observe what index is used at run-time. See the Dan Foremanís PPTG book for more info (Progerss msg 1848).