You can download either the 32-bit or 64-bit version of UCINET. The 32-bit version is the standard version and runs on both 32-bit and 64-bit Windows. The 64-bit version is limited in that it does not have all of the functions of the 32-bit version. It often crashes. Therefore, it is best used in tandem with the 32-bit version.

32-bit Installation Package. This installs the 32-bit version of UCINET along with several helper programs (such as NetDraw), and puts a copy of all the standard datasets in a folder called Ucinet Data under your Documents folder. It runs on 64-bit and 32-bit Windows. The installation program is new as of version 6.531 and unfortunately can't automatically uninstall versions of UCINET prior to 6.531, so you should do that yourself prior to installing current version.

64-bit installation package. This is NOT recommended. If you have a 64-bit version of Windows, you can try this 64-bit version of UCINET. It lets you analyze much larger datasets, if you have the RAM memory for it. However, it is also flaky. Older parts of UCINET are not compatible with 64-bit execution, so these parts will crash. 

Trial vs "real" version. Actually, there is no separate trial version. If you download the program, it will run on your computer for 60 days without having to enter a registration code. Buying the program gets you that registration code. So whether you buy or just try, you download the same program.

Virus/malware issues. There was a time when the download files were stored in a DropBox location. For some reason, this caused Norton AntiVirus to flag the program as dangerous. So we don't use DropBox for this anymore. But you might still run into issues with Norton or some other anti-virus program when installing UCINET. One thing to be aware of is that Windows 8 will issue a warning that the publisher is unknown and probably dangerous. This just means we haven't sought (read: paid for) Microsoft certification.

Installation Notes


Installation works best if you right-click on the installation file (typically called something like setup32UCI6534.exe) and choose 'Run as Administrator'. Similarly, it is also helpful to use 'Run as Administrator' when running UCINET itself for the first time, as this allows you to register the program for all users of your machine. See the FAQ on this.


The best way to run UCINET on a Mac is to use a Windows emulator such as Parallels (or, of course, Bootcamp). However, it is (often) possible to run UCINET on a Mac using Wine. For more information, see this FAQ

Version Info

  • Version 6.627 | 14 Dec 2016
    • Removed in limitation in Network|Subgroups|Clique that was preventing the production of the clique-clique overlap matrix when the number of cliques was greater than 255. The new limit is 5000. 

    Posted Dec 14, 2016, 2:40 PM by Steve Borgatti
  • Version 6.626 | 13 Dec 2016
    • Fixed bug in Network|Cohesion|E-I Index. It was screwing up the attribute vector when there were zeros present.

    Posted Dec 13, 2016, 1:57 PM by Steve Borgatti
  • Version 6.625 | 16 Nov 2016
    • Fixed bug in the DL editor. When importing data in matrix form in which there were duplicate labels (e.g., two nodes named "steve"), the program was incorrectly assigning all ties to the first of the duplicate nodes. 
    Posted Nov 16, 2016, 7:04 PM by Steve Borgatti
  • Version 6.624 | 15 Oct 2016
    • Added subgraph command to the CLI. Given a square matrix representing a network and a vector indicating membership of nodes in groups, the subgraph command allows you to extract a subgraph consisting only of nodes belonging to one or more groups. The syntax is <outfile> = subgraph(<input matrix> <attribute vector> <relational operator> <value>). For example:
      ->gender = cols(campattr 1)
      ->boys = subgraph(campnet gender eq 1)
      ->girls = subgraph(campnet gender eq 1)

      If the operator and value are omitted, the program assumes "GT" and "0". For example, if you have a vector of 1s and 0s indicating who is a member of a given component, you could write:

      ->mainweb = subgraph(campnet maincomp)

      This would select all nodes with a value greater than zero on maincomp.

    Posted Oct 16, 2016, 5:30 AM by Steve Borgatti
  • Version 6.622 | 8 Sept 2016
    • Fixed discrepancy in how Networks|Compare Densities|Paired and Networks|Compare Densities|Theoretical parameter worked. Now, given the same random seed, they produce the same density standard errors

    Posted Sep 11, 2016, 11:46 AM by Steve Borgatti
  • Version 6.621 | 5 Sept 2016
    • Added cosine similarity (aka Tucker's congruence coefficient) to Tools|Similarities & Differences. 
    • Fixed bug in matrix editor which was adding rows and columns gratuitously
    Posted Sep 5, 2016, 3:19 PM by Steve Borgatti
  • Version 6.620 | 5 July, 2016
    • Added exact keyplayer routine to the CLI. The procedure finds the best set of k nodes which, if removed from the network, would maximize the extent of fragmentation (as measured by the number of pairs of nodes that can't reach each other by any path). The algorithm is brute force: it checks all possible subsets of size k. As a result, it is only appropriate for small networks and small k. For example, it can solve N=50 and K=3 in a few seconds, but N=100 and K=3 takes several minutes. Syntax is exhaustivekp(<dataset> <k>), such as:

      ->dsp exhaustivekp(zackar 3)

    • Installer package contains version 2.160 of NetDraw, which changes the way the Width checkbox on the Rels tab works. Before, the width of the line between two nodes was determined by the average tie strength of all tie types currently active. Now, the width is determined by the largest tie strength of any currently active types of tie between that pair of nodes. 
    Posted Jul 5, 2016, 9:32 AM by Steve Borgatti
  • UCINET 6.620 | 13 June, 2016
    • Fixed important bug in Data|Affiliations|Mode=Rows. When given a multi-matrix dataset as input, it was writing out each 1-mode projection twice.
    • Replaced k-plex routine in Network|Groups|K-Plexes
    Posted Jun 13, 2016, 11:12 AM by Steve Borgatti
Showing posts 1 - 8 of 210. View more »