Summary Despite national and international regulations requiring protection for workers at risk of falling, falls from heights are still a primary cause of death for construction workers. In fact, they are the second leading cause of accidents in terms of costs ($397 million a year for 2010 to 2012) and represented 16.1% of workplace fatalities in 2017. Temporary guardrails are an effective way to protect workers against falls from heights; being a passive means of protection, they do not interfere with worker productivity. The design of the vertical posts and railings of temporary metal guardrails does not pose any major problem and was studied in an earlier research project conducted by the Institut de recherche Robert-Sauvé en santé et en sécurité du travail (IRSST) (Lan and Daigle, 2011). However, the strength of a guardrail system depends in part on how it is anchored to the support structure. Guardrail baseplates are usually nailed or screwed to the support structure. There have been no studies of wood guardrails installed on open-web joists or prefabricated walls (during building construction) to date, so it is hard to know their resistance to the loadings prescribed by the Safety Code for the Construction Industry (SCCI) in force in Quebec. For the metal guardrails generally used for waterproofing work on existing buildings, the strength depends on the fastening base, which the contractor often knows little about. The strength of metal guardrails can be estimated based on empirical formulas modelling the pull-out strength of the screws. However, the variability of the pull-out strength determined by these empirical formulas can be significant. Furthermore, the in situ conditions of the wood are very often unknown (type of wood, moisture, degree of rot, presence of knots). The strength of guardrail attachment to real structures is therefore very often approximate. The goals of the research were to (i) determine the strength of wood guardrails fastened to new open-web joist structures erected in the laboratory, (ii) determine the strength of wood guardrails fastened to prefabricated walls erected in the lab and (iii) compare the behaviour under load of metal guardrails installed on flat roofs of existing structures of various ages. In the study, wood guardrails built with 2” x 4” lumber and fastened to open-web joists underwent 262 resistance tests in the lab. With these tests, the researchers were able to analyse the influence of the following variables: guardrail height (1 m or 1.2 m), joist height (9.5 in., 12 in., 14 in., 16 in.), test configuration (1 span, 3 spans, 2 spans with force applied directly to the vertical post) and types of securement (using smooth nails, ring shank nails, screw shank nails, wood screws, lag screws). The influence of these parameters was also studied in the lab when 98 resistance tests were run on wood guardrails attached to a prefabricated wall built out of 2” x 4” lumber. In these tests on open-web joists and a prefabricated wall, horizontal force was applied to the top plate of the guardrail using a manual winch and measured by means of a load cell and a data acquisition system with a frequency of 10 Hz. The vertical load was a dead weight applied to the top plate of the guardrail. Last, 36 tests were conducted in the field with metal guardrails installed on real structures: two buildings (one from 2008 and the other from 2013) and four different parapets (different dimensions and types of construction), all in good condition before the testing (moisture content of less than 10% and no visible damage). Three models of metal guardrail were tested, including two that had already been investigated in the lab in an earlier study. For these guardrails, several different means of attachment to parapets were studied: securement on one or two sides (with a parapet clamp or stabilizing plate) and with various types of screws (black screws, self-tapping screws and lag screws). The results of the laboratory tests showed that the resistances obtained were greater for 1‑m‑high guardrails and that the most critical test configuration was the one where force is applied directly to the vertical post. Both of these results were expected. Generally speaking, the resistances recorded were greater for larger joists, while the attachment height did not seem to have a notable influence for the prefabricated wall. In the latter case, the results observed were probably due to the lack of depth of the nailing base (less than an inch). Last, the results also showed that some types of attachment offer better resistance: when the vertical post is loaded about its strong axis (which is not as easy as it looks to put into practice in reality), when wood or lag screws are used, or when a metal brace is used along with smooth nails to strengthen the attachment of the post to the support structure. The results of the field tests showed that the type of screws used to anchor the guardrail bases has a certain influence, but the total number of screws used in all remains the most important variable. It is essential to follow the manufacturer’s instructions regarding the number of screws to use. In the tests, none of the guardrails attached to just one side of the parapet achieved the resistance required by the SCCI. For small parapets consisting of an insulating panel that extends above the level of the roof (which represents the most critical case), when the post is attached on two sides, the resistances obtained did meet SCCI requirements. As some types of guardrails are hard to install on small parapets, it is better to choose a model of guardrail that has modular mounting plates that can be adapted to the greatest number of possible situations.