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 6.620 | 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
  • UCINET Version 6.619 | 10 June 2016
    • Made change in CLI to force use of courier new font. Previously, it was using the default Windows system font, which was almost always courier, but occasionally not. Using courier ensures that columns of numbers line up properly
    [ download
    Posted Jun 10, 2016, 4:51 AM by Steve Borgatti
  • Version 6.618 | 4 June 2016
    • Added new command Data|Filter/extract|Extract Columns. Use it to pull out one or more columns from a dataset and put them in a new dataset.
    Posted Jun 5, 2016, 9:24 AM by Steve Borgatti
  • Version 6.617 | 3 June 2016
    • Fixed bug in Tools|Testing hypotheses|Node level|Regression that was causing the program to always choose the first column of the dependent variable dataset
    • Fixed bug in the matrix editor which was causing crashes on pressing the New button.
    Posted Jun 3, 2016, 3:40 PM by Steve Borgatti
  • Version 6.616 | 30 May 2016
    • Fixed bug in the CLI's Forfiles command. Previously, commands of the form ->forfiles samplk? draw were failing, while commands like ->forfiles samplk* $-cent = centrality($) were succeeding.
    • Note: bug discovered in this version in Tools|Testing hypotheses|Node-level|Regression. No matter which column you select as dependent variable, it always runs the regression on column 1

    Posted Jun 1, 2016, 5:45 PM by Steve Borgatti
  • Version 6.615 | 29 May 2016
    • Note: bug discovered in this version in Tools|Testing hypotheses|Node-level|Regression. No matter which column you select as dependent variable, it always runs the regression on column 1
    • Added "number of holes" statistic to CLI's Holes routine. It simply counts the number of pair of alters that are not connected. This captures both network size and density
    • Added limited Frequencies command to the CLI. Usage: 
    ->dsp freq(sampson)

    Matrix: Frequencies

              1   2   3   4   5   6   7   8   9  10 
              1   2   3   K       S       N       R 
            --- --- --- --- --- --- --- --- --- --- 
        1 0 251 249 250 259 252 248 253 256 267 265 
        2 3  18  18  18  15  17  16  17  16  13  13 
        3 1  18  21  19  15  18  20  17  16  14  14 
        4 2  19  18  19  17  19  22  19  18  12  14 

    Matrix: Proportions

                1     2     3     4     5     6     7     8     9    10 
               K1    K2    K3    LK     S    ES     N    IN     R    PR 
            ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- 
        1 0 0.820 0.814 0.817 0.846 0.824 0.810 0.827 0.837 0.873 0.866 
        2 3 0.059 0.059 0.059 0.049 0.056 0.052 0.056 0.052 0.042 0.042 
        3 1 0.059 0.069 0.062 0.049 0.059 0.065 0.056 0.052 0.046 0.046 
        4 2 0.062 0.059 0.062 0.056 0.062 0.072 0.062 0.059 0.039 0.046 

    Posted Jun 1, 2016, 2:47 PM by Steve Borgatti
  • Version 6.614 | 22 May 2016
    • Note: bug discovered in this version in Tools|Testing hypotheses|Node-level|Regression. No matter which column you select as dependent variable, it always runs the regression on column 1
    • Changed Network|Compare aggregate proximity matrices|partition to be able to handle missing values
    • Changed the CLI's IPF routine to default to treating diagonal values as valid. If you want the diagonal ignored you must add the word IGNORE to the command. E.g.,
      ->dsp ipf(padgm ignore)

    Posted Jun 1, 2016, 2:48 PM by Steve Borgatti
Showing posts 1 - 8 of 204. View more »