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. 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. 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 6.596 | 22 Jan 2016
    • Changed netdraw button on toolbar to immediately ask for a file name before starting netdraw
    • Added cutpoints routine to CLI (aka matrix algebra). Given a network, it outputs a vector V in which v(i) = 1 if node i is a cutpoint. Unlike menu version, works for both undirected and directed graphs. Syntax and example are as follows:

      -><outfile> = cutpoints(<dataset>)

      ->cp = cut(campnet)
      ->dsp cp

    Posted Jan 22, 2016, 6:21 PM by Steve Borgatti
  • Version 6.595 | 16 Jan 2016
    • Fixed bug in Double Dekker QAP regression. When coefficients were sufficiently small, the program was setting them to zero. As a result, when it compared the observed coefficient (zero) with the coefficient from a random permutation (also zero), it was always finding that the permutation value was as large as the observed, yielding a p-value of 1.0.   
    • Fixed bug in Transform|Symmetrize which was only symmetrizing the first matrix in a dataset
    Posted Jan 16, 2016, 12:35 PM by Steve Borgatti
  • Version 6.594 | 16 Dec 2015
    • Yet another install package for the 64-bit version -- previously, after installation the program would not start up
    • Change in Networks|Cohesion|Alters in Common so that valued data would automatically be dichotomized. In addition, missing values are now ignored
    • Bug fixed in Transform|Aggregate|CSS. Previously, the labels Pa and pA were reversed. Now, the value labeled Pa correctly gives the number of ties Perceived but not Actually there, while pA gives the number of Actual ties that the respondent failed to Perceive
    Posted Dec 16, 2015, 10:32 AM by Steve Borgatti
  • Version 6.593 | 9 Dec 2015
    • Uploaded new install package for the 64-bit version
    • Added option to Gould & Fernandez brokerage to allow closed triads to count as brokerage instances as well as open triads
    • Reverted minres routine in continuous core-periphery procedure to old version that seems to handle negative values and non-symmetric data better 
    Posted Dec 9, 2015, 12:22 PM by Steve Borgatti
  • Version 6.590 | 12 Nov 2015
    • Minor change in Transform|Aggregate|CSS to make default output file names match the method of creating the "true" network
    • Fixed bug in Networks|Cohesion|Density|Density by Groups which was causing crashes when the user selected the Identity partition instead of a categorical attribute
    • Fixed another bug in Networks|Cohesion|Density|Density by Groups which was failing to retain node labels when Identity partition was used
    • Added CLI command for aggregating columns of a matrix, much like Density by Groups. It is called AggCol and is used like this:

      ->gender = col(campattr "gender")
      ->dsp aggcol(campnet gender)

      The result, in this case, would be counts of the number of male and female alters that each node has:

               1 2 
               1 2 
               - - 
          1  1 2 1 
          2  2 0 3 
          3  3 3 0 
          4  4 3 0 
          5  5 3 0 
          6  6 3 0 
          7  7 3 0 
          8  8 3 0 
          9  9 1 2 
         10 10 0 3 
         11 11 1 2 
         12 12 1 2 
         13 13 1 2 
         14 14 1 2 
         15 15 0 3 
         16 16 0 3 
         17 17 0 3 
         18 18 0 3 

    Posted Nov 13, 2015, 10:32 AM by Steve Borgatti
  • Version 6.588 | 25 Oct 2015
    • Added routine in CLI to calculate Goodman and Kruskal Gamma coefficients (measures of rank-order similarity) on columns of a matrix. Example usage is:

      ->coefs = ordinalassoc(campnet)
      ->dsp coefs
    Posted Oct 25, 2015, 6:15 PM by Steve Borgatti
  • Version 6.587 | 6 Sept 2015

    • Added att2mat command to CLI. This converts an attribute to a matrix. The syntax is <outfile> = att2mat(<vector> [<method>]) where the choices for method are: 'match', 'diff', 'absdiff', 'sqrdiff', 'prod', 'sum', 'identity', 'rows', and 'cols'. The default is 'match'. Example:
      ->gen = col(campattr gender)
      ->samegender = att2mat(gen match)
    • Added mat2vec command to CLI. This vectorizes a matrix so that it looks like one long column. If the input dataset contains multiple relations, then each becomes a column in the vectorized dataset. The syntax is <outfile> = mat2vec(<dataset> [<method>] [diagok]) where the choices for method are: 'whole', 'lowerhalf' and 'upperhalf'. The default is 'whole'. By default, the diagonal is omitted, but if want the diagonal included, add 'diagok' to the list of parameters. Example:
      ->pvec = mat2vec(padgett lowerhalf)
      ->svec = mat2vec(sampson whole diagok)
    Posted Sep 6, 2015, 8:57 AM by Steve Borgatti
  • Version 6.586 | 23 July 2015

    • Fixed bug in Matrix editor. It had been crashing when the user saved a dataset after having cleared the editor using the New button
    Posted Jul 23, 2015, 6:33 PM by Steve Borgatti
Showing posts 1 - 8 of 182. View more »